Beyond the Tip of the Iceberg: Rethinking Hypertension Screening in Young Adults

In 2023, >1 in 5 young adults aged 18 to 39 years in the United States had hypertension, yet of them, only 28.3% were aware of their condition and just 5.6% achieved blood pressure (BP) control to <130/80 mm Hg. ¹ Within this age cohort, young adult men and Black or Hispanic individuals had even lower rates of awareness, treatment, and control of hypertension, ¹ a phenomenon that has been observed in other developed countries, such as the United Kingdom. ² Although estimates vary depending on diagnostic thresholds, the prevalence of hypertension worldwide shows that as many as 1 in 8 young adults may be affected. ³ , ⁴ Alarmingly, only a small fraction of young adults with hypertension are aware of their condition and are receiving treatment, just the proverbial tip of the iceberg. This is concerning, as hypertension is a major risk factor for cardiovascular disease (CVD). Although overall CVD rates have declined in the United States in recent decades, the proportion of CVD events among young adults is increasing. In fact, cardiovascular mortality rates are increasing in certain demographic groups, rapidly. ⁵ , ⁶ Similar trends are seen on the global scale, with significant increases in the prevalence of CVD among young adults. ⁴ Among increases in other cardiovascular risk factors, such as obesity and diabetes, ⁷ hypertension remains a leading risk factor for CVD in young adults. ⁸ Early identification and treatment are vital to reducing lifetime cardiovascular risk. ⁹ , ¹⁰

Among young adults, we now understand from various studies that several independent risk factors, such as obesity, pregnancy‐associated hypertension disorders, a family history of hypertension, and other cardiovascular morbidities, such as diabetes or dyslipidemia, are associated with a greater risk of developing hypertension. ¹¹ However translation of this knowledge to a clinical application that is able to provide a robust screening pathway for young adults that can be acted upon has been limited in the absence of a credible risk score. ¹² The US Preventive Services Task Force recommends annual BP screening for adults aged ≥40 years, as well as for those aged 18 to 39 years who are at increased risk for hypertension, including Black individuals, those with high‐normal BP, and individuals who are overweight or obese. For adults aged 18 to 39 years who are not at increased risk and have previously normal BP readings, screening is recommended every 3 to 5 years. ¹³ This guidance, however, is at odds with other developed countries, such as the United Kingdom, where screening is not routinely recommended for hypertension in those aged <40 years. Because of in part the poor specificity, in rationalized and economically constrained health care systems, it is not cost‐effective to measure everyone's BP so screening programs for hypertension in young adults are not yet judged to be a good use of resources. This has meant that case finding of asymptomatic hypertension, the “silent killer,” ³ in young adults remains inconsistent and sporadic. It follows that to make a viable hypertension screening program, we must first address the unmet need for a hypertension risk prediction model for young adults that is able to identify those at highest risk of developing hypertension who would benefit most from targeted screening.

In this issue of the Journal of the American Heart Association (JAHA), An et al ¹⁴ have developed and validated an incident hypertension risk prediction model for young adults that was derived and internally validated from a diverse contemporary cohort of 355 524 young adults aged 18 to 39 years from Kaiser Permanente Southern California and externally validated on 3 other cohort studies. Their model incorporates predictors, including age, systolic and diastolic BP, body mass index, smoking, dyslipidemia, high‐ and low‐density lipoprotein cholesterol, diabetes, and hypertensive pregnancy disorders. The model demonstrates strong discrimination and calibration. Compared with the US Preventive Services Task Force recommendations, which estimate that 45.3 million young people in the United States are at risk for hypertension, this risk prediction model offers improved specificity, without compromising sensitivity, reducing the estimated at‐risk population to 28.5 million. The model also gives insight into the relative weight of risk factors for developing hypertension in young adults as it is predominantly driven by obesity, diabetes, and previous BP readings (in men and women) and hypertensive disorders of pregnancy (in women only).

There are limitations to the model in its current form, which in the most part stem from limitations of the cohort data used to develop and validate the model. First, the Kaiser Permanente Southern California cohort includes insured young adults only, which may lead to underrepresentation of those with lower socioeconomic status and those who do not present to health care facilities. This is mitigated to some extent as the model is externally validated in more representative cohorts of young adults in the community. Second, the cohorts used are all United States based and at a regional level demonstrate ethnic preponderance, such as the plurality of Hispanic Americans in the Kaiser Permanente Southern California cohort that brings into question the generalizability of such a model to other countries with different ethnic make‐up. The authors also submit that additional hypertension risk factors, such as diet, physical activity, and family history of hypertension, were explored but not included in the final model because of a paucity of data. Regardless, a key variable that is included in the model, baseline BP measurement, is not readily available in young adults outside of cohort studies within the general population, which brings into question the utility of such a risk prediction model in its current form.

We propose how the work of An et al and the existent US Preventive Services Task Force recommendations can be adapted for routine practice. Using a hypertension risk prediction model such as that developed by An et al, we offer an example of how a potential real‐world community hypertension screening program for young adults could function.

The Figure suggests a potential pathway whereby patients identified from primary care (community practice) health records who are aged <40 years with ≥1 of obesity, hypertensive disorder of pregnancy, diabetes, dyslipidemia, family history of hypertension, or a smoking history are either invited to submit an out‐of‐office BP measurement or attend for an office BP measurement. These BP measurements are then entered into the prediction tool to stratify their risk, which informs the clinician whether they should have ongoing monitoring and at what interval. Although there remains questions on the accuracy and performance of cuffless wearable BP devices, ¹⁵ it is likely that market‐leading smartwatches, which have seen mass adoption in the past decade, may soon be able to record and store BP measurements, which could influence and inform these models. It is then conceivable that an individual with a wearable device, be it a smartwatch or one of the emerging smart rings, identified to have 1 of the aforementioned risk factors for developing hypertension, could upload their average BP readings from the device remotely for input into a risk prediction tool, and the system may provide an alert and automatically lead to review by a clinician.

Figure 1. Flowchart demonstrating how a hypertension risk prediction model could be incorporated into a hypertension screening program for young adults.

*Alternatively, a patient could be invited to submit an out‐of‐office BP measurement from either a traditional cuff‐based device or a wearable cuffless device. If BP was raised, they could then be offered a confirmatory office BP measurement. BMI indicates body mass index; and BP, blood pressure.

Although most young adults with hypertension have a low absolute risk of CVD, a minority will develop severe and potentially fatal complications early in life, such as renal failure, myocardial infarction, and stroke. Studies looking at how we can predict such outcomes in this group of patients specifically are needed. However, early detection presents benefits for all young adults with hypertension. Secondary causes of hypertension, such as primary hyperaldosteronism and renal artery stenosis, that are estimated to account for ~10% of cases of hypertension in young adults can be diagnosed earlier and potentially cured, which in some cases can prevent the need for long‐term pharmacologic antihypertensive therapy. Although absolute risk of CVD from hypertension is low for a young adult, the cumulative lifetime risk of hypertension for various types of CVD is high. ¹⁶ Additionally, we now understand that left untreated, chronic high BP can lead to permanent structural changes in the arterial wall, which increases arterial stiffness and systolic BP, which can drive further vascular changes and increases in BP in an insidious feedback loop. ¹⁰ , ¹⁷ Consensus guidelines are now reflecting our greater understanding of the pathophysiology and risk of hypertension in young adults as they move toward lower BP treatment thresholds and recognize that some patients at increased risk of CVD with “elevated BP” (office systolic BP 120–139 mm Hg or diastolic BP 70–89 mm Hg) may benefit from BP‐lowering therapy. ¹⁸ , ¹⁹

In summary, An et al have developed a novel risk prediction model that moves us closer to understanding how best to case‐find young adults with hypertension to enable early intervention and treatment and reduce cardiovascular risk. We have proposed a framework for 1 clinical application that incorporates a prediction model such as this, but there are many other options and use cases. There is of course need for further research and development, as is outlined in the Table, that will be necessary to bring a hypertension risk prediction model into public health systems and clinical practice. The utmost caution must, however, be exercised as machine learning algorithms and artificial intelligence become ever‐more common adjuncts in health care risk prediction and decision‐making. Concerns around algorithmic bias are not unfounded and must be taken seriously and guarded against when designing and appraising these tools. ²⁰ This marks a crucial step forward in addressing a major unmet need. For hypertension screening to be effective, affordable, and widely accepted, we must be able to accurately and cost‐effectively/efficiently identify those at greatest risk. This risk prediction model shows how that can be done, sharpening our focus beyond the tip of the iceberg. With its potential to significantly reduce long‐term cardiovascular risk, not just in young people but across the broader population, it paves the way for meaningful, lasting public health impact.

Table 1.

Outline of Key Steps That Need To Be Taken Before a Hypertension Risk Prediction Model Is Ready To Be Adopted in Health Care Systems

Where next…

Further studies are required to develop and improve prediction models

Develop prediction models and validate with different populations from different countries

Further assessments of efficacy

Incorporate more risk factors, including social determinants of health, family history, diet, and physical activity, into future models

Further understanding of the validity of office vs out‐of‐office BP measurements and cuffless devices

Different countries may need to adapt different models because of different risk profiles

Cost‐effectiveness and health economic analysis

Trial of prediction models incorporated into real‐life health care systems

BP indicates blood pressure.

Disclosures

None.