Hypertension is a major cardiovascular risk factor associated with a two- to three-fold increase in risk of developing atherosclerotic cardiovascular disease.1 Conversely, randomized trials have shown that reducing blood pressure (BP) significantly reduces the risk of cardiovascular disease.2

The prevalence of hypertension in Europe is 30-45%, higher in older age-groups.3 In Portugal, the PHYSA study carried out by the Portuguese Society of Hypertension in 2011-2012 revealed an overall prevalence of 42.2% in the adult population.4 In a 2013 study in a primary health care (PHC) setting, the prevalence was 29.1%.5

Prevention, diagnosis and treatment of hypertension mainly takes place in PHC settings, where most people are managed.6 Given its high prevalence and associated morbidity and mortality, hypertension represents a considerable burden on the health system, requiring regular monitoring and prescription of drug and/or non-drug therapies, which consume large quantities of resources. Accurate diagnosis is thus important.

The most common method for diagnosing hypertension is office BP (OBP) measurement, but this has certain limitations. Population-based studies show that the prevalence of masked hypertension (elevated home or ambulatory BP but normal OBP) is around 13%. White coat hypertension (WCH) (office but not home or ambulatory BP above cutoffs for hypertension3) has a similar prevalence and accounts for 32% of diagnoses of hypertension,7 which highlights the low sensitivity and specificity of the standard measurement method. The white coat effect is defined as OBP ≥20/10 mmHg higher than their home or ambulatory BP.8 Patients with WCH incur unnecessary financial and social costs, while those with masked hypertension may suffer by remaining untreated.

Ambulatory blood pressure monitoring (ABPM) has greater reproducibility than OBP measurement (OBPM) and correlates better with target organ damage.9,10 ABPM is used when there is uncertainty regarding the diagnosis of hypertension, treatment resistance, irregular BP or suspicion of WCH.11 The method measures BP several times over the course of 24 hours. It is considered the best exam for diagnosis of hypertension, the mean 24-hour value being associated with better risk prediction than OBPM,3 as well as identifying BP patterns that indicate higher risk of cardiovascular events and target organ damage such as non-dipper pattern (defined as a fall of <10% in nighttime BP12) and masked hypertension, in different subpopulations.12–17 However, its use is limited by the need for qualified technicians and its high cost.18 Despite these disadvantages, ABPM is increasingly recommended in clinical guidelines: in 2001 the US Center for Medicare and Medicaid Services approved ABPM for identification of WCH,19 and since 2011 the UK National Institute for Health and Care Excellence (NICE) has recommended it for all PHC users with BP >140/90 mmHg as a cost-effective way of diagnosing hypertension.8

The aim of this study was to review the cost-effectiveness of ABPM compared to other methods in the management of hypertension.

We performed a literature search in the Canadian Medical Association Infobase, Guidelines Finder, National Guideline Clearinghouse, Bandolier, BMJ Clinical Evidence, the Cochrane Library, the Database of Abstracts of Reviews of Effects, Medline, the Trip Database, SUMSearch and Índex das Revistas Médicas Portuguesas. We researched articles published between January 2005 and August 2015 in Portuguese, English and Spanish. Each author performed an independent search during September 2015 and the data were subsequently cross-checked and replicated. The search was updated up to February 29, 2016 before this review was submitted.

The publications searched were meta-analyses, systematic reviews, clinical trials, observational studies and clinical guidelines. Other publications in the reference lists of selected publications were also selected when considered relevant.

We used the MeSH terms “Hypertension”, “Blood Pressure Monitoring, Ambulatory” and “Cost-Benefit Analysis” and the Portuguese search terms (Descritores em Ciências da Saúde) “Hipertensão”, “Monitorização Ambulatorial da Pressão Arterial” and “Análise Custo-Benefício”. Since using a single query combining all these terms simultaneously would have severely limited the search results, several searches were performed using different combinations of the terms.

The inclusion criteria used were the following:

• Population: individuals aged ≥18 years

• Intervention: 24-hour ABPM

• Comparison: other methods of BP measurement (home BP monitoring [HBPM] and/or OBPM)

• Results: cost-effectiveness for diagnosis and/or management of hypertension

Studies of pregnant women, hospitalized or immunocompromised individuals, and those with acute disease, very high cardiovascular risk or secondary hypertension were excluded, as were consensus opinions and articles of dubious methodological rigor.

The selection process consisted of the following steps: determination of the initial total number of articles; exclusion of repeats; exclusion based on reading of titles and abstracts revealing presence of exclusion criteria or absence of inclusion criteria; reading of the remaining articles in their entirety and selection on the basis of the inclusion and exclusion criteria; selection of articles to include; and re-reading of selected articles in their entirety. They were then classified according to the study design. Levels of evidence and grades of recommendation were attributed according to the Oxford Centre for Evidence-Based Medicine scale.20,21 The grades of recommendation attributed by the studies themselves were respected, irrespective of the levels of evidence used, and were specified when applicable.

A total of 525 articles were initially found. Of these, 43 were excluded as repeats and 443 were excluded after reading of titles and abstracts for not fulfilling the inclusion criteria or fulfilling at least one exclusion criterion. Of the remaining 39 articles read in their entirety, 33 were excluded for falling outside the scope of the review or failing to meet the above criteria, mostly due to the lack of a robust analysis of cost-effectiveness. The final selection was thus of six articles, five original studies and one clinical practice guideline. No publications found in the reference lists of the included studies were added.

Since these articles are relatively heterogeneous, they are presented and discussed individually. The original studies are summarized in Table 1.

The study by Rodriguez-Roca et al. assessed the cost-effectiveness of ABPM compared to OBPM in the management of patients with hypertension.22 This was an observational study of a convenience sample of 241 individuals in a rural Spanish setting diagnosed with hypertension by OBPM and managed in PHC. The measure of effectiveness of treatment was the proportion of cases with controlled BP, defined as OBP lower than 140/90 mmHg (<130/85 mmHg in patients with diabetes, target organ damage or cardiovascular disease) or mean daytime BP <135/85 mmHg by 24-hour ABPM.

The cost-effectiveness analysis focused on the direct cost of follow-up for one year by OBPM or ABPM per controlled patient. Costs were calculated for two strategies: follow-up every three months by OBPM for one year, or OBPM and initial ABPM followed by OBPM every three months for one year. Since both methods were used to measure the BP of all patients in the study (at different times), there was no intervention or placebo group, and both strategies had the same total number of analyzed individuals, the absolute number of well-controlled cases for each strategy was sufficient for comparing effectiveness.

All participants completed the study and there were no losses to follow-up. The costs of the OBPM-only strategy included the cost of assessment by sphygmomanometer and accessories, four nurse visits and two medical consultations per year, and medication costs (of the three most prescribed drugs). For the ABPM strategy, the costs of the equipment, two medical consultations and two nurse visits were added.

This study showed that ABPM was much more effective than OBPM alone for BP control (55.60% vs. 8.29%), irrespective of the antihypertensive treatment used. Cost analysis was also more favorable for ABPM: the mean cost of a well-controlled case using ABPM was around a quarter of the cost by OBPM alone (€238 vs. €940). The analysis showed an incremental cost of €115 per uncontrolled patient managed by OBPM, as well as a saving of €68883 if ABPM was performed in all the patients included in the study from the beginning, obtained by adding the incremental cost to the costs associated with unnecessary treatment for patients apparently uncontrolled on the basis of OBPM only (WCH). The cost of routine initial ABPM as the first-line method for BP measurement was recouped within a year. Although ABPM is initially more expensive, in the long term its cost-effectiveness ratio is more favorable (level of evidence 2).

The study by Krakoff et al. aimed to estimate the cost savings deriving from the use of ABPM, based on an analytical model.23 It compared two homogeneous groups in the US, each with 1000 individuals recently diagnosed with hypertension by OBPM, using annual OBPM in one group and initial ABPM to confirm the diagnosis followed by annual ABPM thereafter in the other, over a five-year period. Those diagnosed with hypertension by ABPM during this period were then monitored only by OBPM. The number of individuals with WCH and those diagnosed with hypertension during follow-up, the number of years and costs of treatment, and the initial and annual costs associated with ABPM were calculated. The analysis was based on data in the literature for WCH prevalence (15-25%), mean cost of ABPM ($75) and of treatment (physician visits, diagnostic tests, and medications), annual incidence of hypertension during follow-up (5-20%), and estimates of annual loss to follow-up and treatment dropout rate (5%). Costs of medication were taken as the lowest (using diuretics), assuming guideline-based treatment.

The results of the study show savings of 3-14% in the cost of treatment of hypertension and a reduction of 10-23% in treatment-years over a five-year period, proportionally to the incidence of hypertension, if ABPM was used when high OBP was recorded. The analysis showed an incremental cost of $156 per year per patient in the group treated using OBPM only. ABPM is cost-effective when annual treatment costs are at least $300, with a total saving of $153019 per 1000 patients in the most favorable scenario.

Whatever the estimate of the annual incidence of hypertension, treatment costs were lower and overall savings were higher with ABPM (level of evidence 2). Its use for a definitive diagnosis of hypertension and identification of WCH in surveillance of individuals with high OBP reduces the costs associated with the disease, to a greater or lesser extent depending on whether the prevalence of WCH is higher or the incidence of hypertension is lower, respectively.

The study by Lovibond et al. was the first to compare the cost-effectiveness of the three main methods for the initial diagnosis of hypertension (OBPM, HBPM and ABPM).24 It used an analytical model to compare the cost-effectiveness of BP monitoring following OBP of >140/90 mmHg by OBPM (monthly over three months) vs. HBPM (over a week) vs. ABPM (over 24 hours). An incremental analysis was performed to compare the three methods. The model considered a hypothetical PHC population aged 40 years or older and risk-factor prevalence equivalent to that of the British population.

The study was based on a Markov model considering three health states (suspected hypertension, diagnosed hypertension and cardiovascular event). A cycle duration of three months was used, to correspond with the typical period required for a diagnosis by OBPM, and the model was run for the equivalent of 60 years, by which time most people in the population would have died.

Risk of coronary heart disease and stroke was calculated on the basis of the Framingham risk equations, and risk reductions with antihypertensive treatment were based on a meta-analysis by Law et al.25 The model took into account the probability of normotensives becoming hypertensive over time and incorporated periodic reassessment of BP every five years. The model also assumed that treatment would not be harmful if applied to normotensives.

The results were expressed in terms of quality-adjusted life years (QALYs), total costs and incremental costs per QALY gained, discounted at a standard annual rate of 3.5%. The time spent in each health state was calculated and by attributing different costs and quality of life weights to each health state, total costs and QALYs were established for each of the diagnostic options. Effectiveness, expressed in QALYs, was calculated using constants associated with cardiovascular events and transitions between health states taken from a systematic review. The sensitivity and specificity of each diagnostic test was based on data from a meta-analysis by Hodgkinson et al.26

Diagnostic costs for each method included cost and maintenance of equipment and consumibles, treatment costs (physician and nurse consultations and annual cost of hypertension treatment based on the price of the most commonly used generic drug and reassessments), and initial and subsequent costs arising from cardiovascular events.

To minimize uncertainty, the model was run 1000 times and the mean costs and mean QALYs were recorded. Sensitivity analyses were performed to test the robustness of the model's assumptions with different values for a range of variables.

ABPM was the most cost-effective strategy for men and women of all ages, resulting in more QALYs for male and female age-groups older than 50 years (level of evidence 2). In terms of costs, HBPM was similar to OBPM, whereas ABPM was cost-saving. In younger age groups, ABPM had greater cost savings but was associated with a small reduction in QALYs; however, it was more cost-effective than OBPM only. ABPM was still the most cost-effective strategy in most cases when a range of variables were modified including diagnostic costs, failure rates, cardiovascular risk and prevalence of true hypertension. However, it ceased to be cost-effective if reassessment was performed annually rather than every five years.

The Ontario Health Technology Assessment Series review aimed to compare the cost-effectiveness of 24-hour ABPM with that of OBPM for diagnosing hypertension following an OBP reading of >140/90 mmHg.27 A secondary aim was to assess the budget impact of 24-hour ABPM versus OBPM for uncomplicated hypertension; like in Portugal, ABPM is not reimbursed, and its cost is similar (Can$70).

The target population was individuals aged 45 years or older with similar cardiovascular risk to the general population in Ontario and a hypertension prevalence of 22.7%. The criteria and diagnostic strategies were similar to those of the other studies. The results, based on a Markov economic model and using a standard 5% annual discount rate, were expressed in terms of costs, QALYs and incremental costs per QALY gained. The model considered four health states: alive with BP monitoring; coronary heart disease (coronary death, myocardial infarction or angina); cerebrovascular disease (ischemic or hemorrhagic stroke and transient ischemic events); and dead, and possible transitions between them. The economic model was based on an annual cycle and costs and outcomes were calculated over patients’ lifetimes.

Assessment of cardiovascular risk and estimates of relative risk were obtained in similar fashion to the study by Lovibond et al. Costs included physician consultations, hospital costs, equipment (unit price, accessories and maintenance), and prescribed medication.

It was assumed that approximately five visits to the physician's office would be needed in the first six months for an accurate diagnosis of hypertension according to OBPM, reduced to three using ABPM. Based on national figures, a saving of approximately Can$51.80 per patient would be obtained for the first year of diagnosis using ABPM.

However, the literature search performed, which used the Grading of Recommendations Assessment, Development and Evaluation Working group (GRADE) system for classifying the quality of evidence, indicated a lack of a beneficial effect of ABPM on non-fatal cardiovascular events (very low quality of evidence) but improved BP control (moderate quality of evidence) and greater probability of discontinuing antihypertensive therapy (low quality of evidence) even for a follow-up period of one year or less. In a mean follow-up of five years, ABPM resulted in a reduction in total combined cardiovascular events compared to OBPM.

ABPM was more cost-effective both for individuals with raised OBP and for annual BP monitoring of all individuals (level of evidence 2). It was estimated that in the population of Ontario approximately Can$19 million would be saved annually over a five-year period (2011-2015) by implementing ABPM monitoring only when an individual presents raised OBP readings.

The study by Pessanha et al. aimed to assess the diagnostic accuracy and cost-benefit ratio of ABPM in newly diagnosed and untreated hypertensive patients attended in PHC.6 The diagnostic criteria were similar to those of the other studies and the sample was characterized in detail (OBP, ABPM data, gender, age, body mass index, smoking status, dyslipidemia, diabetes and family history of cardiovascular disease).

The study population consisted of 336 subjects selected over a period of 16 months. Initial BP readings were taken in the left arm using a validated oscillometric sphygmomanometer after resting for 10 min and repeated 8-15 days thereafter. To minimize possible classification bias arising from false suspicion of hypertension due to mistakenly recording only a single BP reading, the mean of three readings, repeated after 1-2 weeks, was used for the analysis. Twenty-four-hour ABPM, measured in the non-dominant arm, was applied in individuals with sustained high BP.

The cost-effectiveness analysis compared the cost of medical management (drug treatment, medical visits and laboratory tests) associated with a diagnosis of hypertension by OBPM and ABPM. Parametric tests were used in a statistical model to estimate the results for a total of 1000 patients diagnosed as hypertensive according to OBPM and followed for two years with and without data from ABPM. After ABPM, 206 individuals (61.3%) were identified as true hypertensives (daytime BP ≥135/85 mmHg) and 130 (38.7%) as having WCH (daytime BP <135/85 mmHg). BP levels, and hence cardiovascular risk, were lower in the latter than in hypertensives on both ABPM and OBPM. The costs avoided in cases of WCH include fewer medical visits, laboratory tests and drug prescriptions per year, was well as reduced iatrogenic risks and direct and indirect costs, although annual assessment by ABPM would be required for patients with WCH.3,6,28

The authors estimate that with ABPM total medical expenses can be reduced by 23% (€157500) per 1000 patients followed for two years in PHC. Although ABPM is initially more expensive than OBPM, due to the cost of the procedure itself, it results in significant savings in resources and treatment costs for patients with WCH, reducing medical visits and laboratory tests by half compared to true hypertensives (level of evidence 2).

The clinical practice guideline included in this review was based on the results of the study by Lovibond et al. discussed above. Even though this study was not completed by the time the guideline was produced,8 a partnership was established between the guideline developers and Lovibond's group.

In 2011, the importance in both economic and health care terms of replacing the traditional approach of OBPM with ABPM or HBPM to confirm a diagnosis of hypertension meant that this change was a priority for NICE.

The guideline was based on a cost-effectiveness analysis using a Markov model with multiple deterministic sensitivity analyses that was run 1000 times, the mean of the results being used. All variables and inferences were based on a comprehensive review of systematic reviews, meta-analyses and data on UK unit costs. The results were expressed in QALYs, costs to the UK national health service and incremental cost per QALY gained.

According to NICE, ABPM should be used to confirm the diagnosis of hypertension if OBP is 140/90 mmHg or higher, and HBPM should be reserved for individuals who are unable to tolerate or refuse to undergo ABPM.

The NICE guideline does not provide grades of recommendation. Since the guideline was based largely on the results of a study24 whose authors collaborated in its preparation, the authors of this review also decided not to attribute grades of recommendation. The guideline states that HBPM or ABPM should be considered for differential diagnosis in cases of suspected WCH. ABPM is cost-effective and, in most cases, is associated with significant savings, as well as bringing clear benefits for patients due to its superior diagnostic accuracy.

Of the three main methods of diagnosing hypertension – OPBM, ABPM and HBPM – OBPM has the lowest initial cost. HBPM has an intermediate cost, but its results are more reproducible and are more closely associated with target organ damage than those of OBPM,29 and its values correlate well with mean BP in normotensives but not in hypertensives.15 A meta-analysis of 20 studies with 5683 patients concluded that neither OBPM nor HBPM should be used as the only diagnostic method, due to their low sensitivity and specificity.26 According to NICE, HBPM should only be used for diagnosis when APBM is not readily available or is poorly tolerated by the patient.8,24 However, neither the European3 nor the US30 guidelines contain this recommendation. Due to its cost, in many countries, including Portugal, ABPM is only used in cases of uncertain diagnosis, suspected WCH or resistance to treatment.11 It is also essential that the equipment should be validated and maintained, in order to ensure the reliability of the readings.31

ABPM minimizes the observation and recording bias inherent to OBPM and HBPM, and thus provides a BP profile that is more reproducible and closer to reality. It is superior at differentiating between true hypertension and WCH, but it is not perfect, and OBPM should still be used to estimate cardiovascular risk.8 WCH is associated with increased risk, albeit less than sustained hypertension.32,33 There is no evidence to support drug therapy in patients with WCH, although lifestyle changes and regular monitoring are recommended.3,6 Detection of WCH by ABPM avoids unnecessary treatment, preserving the quality of life of these individuals.34

ABPM is also useful in the opposite case, that of masked hypertension. This is associated with increased cardiovascular risk, closer to that of diagnosed hypertension than of WCH, and may require intervention and/or regular surveillance.35,36

The additional benefits of ABPM, not all covered in this review, include greater specificity in diagnosis (fewer false positives), lower expenditure on medication for patients and for the health system, better BP control, assessment of the impact of starting and/or modifying treatment and reducing the labeling effect (the negative psychological effect of a diagnosis of hypertension). Its greater sensitivity means it can diagnose cases that OBPM cannot, and so patients at greater cardiovascular risk do not suffer by remaining untreated. It also reduces the time required to establish the diagnosis, so physician visits can be scheduled earlier. However, we found no studies that quantify the effect of such delays or their impact on the course of the disease, particularly in terms of morbidity and mortality.

The studies analyzed in this review conclude that ABPM is cost-effective when used to confirm an initial diagnosis of hypertension. According to Lovibond et al. it is the most cost-effective method as well as leading to more QALYs for patients of both sexes aged over 50 and is especially advantageous for reassessment at five-year intervals of individuals with WCH.24 The Ontario study group found ABPM to be cost-effective for annual BP monitoring of all individuals, although less so than when used only following high OBP readings.27 For Rodriguez-Roca et al., the cost of a well-controlled patient using ABPM is a quarter of that using OBPM, while its effectiveness for BP control is around seven times greater.22 According to Krakoff et al. ABPM is cost-effective, leading to savings of 3-14%,23 and Pessanha et al.’s study revealed savings of 23% over a two-year follow-up.6

The studies included have certain limitations. None analyze the indirect costs associated with a diagnosis of hypertension, such as absenteeism or social and psychological effects, and therefore potentially underestimate the advantages of ABPM.

The study by Rodriguez-Roca et al. was based on a convenience sample with specific characteristics, which limits its conclusions to similar populations only. The high prevalence of cardiovascular risk factors and older mean age could have led to high BP levels on OBPM, enhancing the benefit of ABPM in distinguishing true hypertension from WCH. It does not compare QALYs, and the methods for calculating the total costs associated with each strategy are unclear, specifying only the annual costs of OBPM equipment and of anti-hypertensive medication. Furthermore, the robustness of the results is weakened by the short follow-up period.

The study by Krakoff et al., based on an analytical model, does not include other comparators, including QALYs, BP control or cardiovascular events, and the composition of the study sample is unclear, as are the sources for prevalence of WCH and annual loss to follow-up. The analysis of costs does not specify discount rates or include monitoring of treatment or of patients with refractory hypertension, but costs appear to be lower with ABPM. Sensitivity analysis was performed for two parameters, prevalence of WCH and incidence of hypertension, which added to the robustness of the results. If not all patients are medicated with the least expensive antihypertensive drug, the benefit of ABPM could be greater.

The study by Lovibond et al. is vague in its characterization of the study sample, the frequency of HBPM measurements, the method of assessing cardiovascular risk (which was not based exclusively on UK data), and the attribution of maximum sensitivity and specificity to ABPM, which were not adjusted for age or health state. However, it does present multiple sensitivity analysis of the variables included and the results are robust. It underestimates the benefit of ABPM since, contrary to the authors’ assumption, not all patients are medicated with generic drugs, and the impact of unnecessary treatment is not accounted for.

In the Ontario study, the prevalence of coronary and cerebrovascular disease and the age distribution of the sample assumed by the authors are taken from studies on UK and US populations and are not necessarily applicable to Canadian populations. The type of anti-hypertensive medication is not stated, and the number of consultations required during follow-up is not specified, although it can be inferred that the effect of ABMP is underestimated.

The study by Pessanha et al. was the only one on a Portuguese population and the only one whose results did not derive mainly from mathematical models. It found a prevalence of WCH of 38.7%, which is in agreement with older but not with more recent studies, and thus may overestimate the benefit of ABPM. In addition, it does not quantify the gains from reductions in iatrogenic effects and in the costs of unnecessary nurse visits, although it is assumed that these would result in additional savings.

Given the small number of original studies on the subject, it is not surprising that we found no systematic reviews or meta-analyses on the cost-effectiveness of ABPM. Although there is firm evidence of the advantages of ABPM for BP control and assessment of cardiovascular risk, there is little evidence on its cost-effectiveness, analysis of which is complex and is mainly performed by means of statistical models. These, however, are less than perfect, and do not include chance or other significant variables. This may explain the lack of studies.37,38

The costs of using ABPM as the preferred method for assessing and controlling BP are also affected by the frequency of the exams. Some authors highlight the need for common sense when deciding how often to perform ABPM, pointing out that annual assessment would not be cost-effective, particularly in younger individuals.24,27 Thus, any recommendation to implement ABPM based on its cost-effectiveness requires evidence on which to base appropriate use, since if excessively frequent, ABPM no longer has economic advantages over conventional methods.

One of the limitations of the present review is the nature of this type of analysis, which is affected by the characteristics of different study populations, variations over time in costs and health policies, socioeconomic factors, and different variables and units used to express the results. Comparisons can thus be problematic, limiting extrapolation to other populations, and so a cost-effectiveness analysis of ABPM for the diagnosis and monitoring of hypertension should ideally be carried out in each country.39

Existing studies have very short follow-up periods, which limits the conclusions to medium-term results. Longer-term studies could be performed, but this is unlikely, given the time required to complete them. Further evidence is needed from studies on Portuguese populations so that in the future ABPM can be more widely available and economically accessible in PHC.

ABPM has greater diagnostic accuracy than OBPM or HBPM, and thus can reduce the adverse reactions associated with unnecessary anti-hypertensive medication and the number of diagnostic exams and of physician and nurse visits involved in BP control, thereby reducing costs. It has been shown that the higher costs of ABPM are outweighed by the savings resulting from its greater diagnostic accuracy and BP control, leading to a better cost-effectiveness ratio. On the basis of the studies analyzed in this review, ABPM as the method of choice for diagnosing hypertension has a grade of recommendation B.

Studies that are not based mainly on estimates or analytical models are needed to compare the three diagnostic methods with sufficiently long follow-up times to determine their effectiveness in terms of morbidity and mortality.

The authors have no conflicts of interest to declare.