STRENGTHS AND LIMITATIONS OF THIS STUDY

Patients with hypertension were recruited from a racially and ethnically diverse population in a real-world clinical environment that has high blood pressure (BP) control rates.

The recruitment strategy leveraged Kaiser Permanente Southern California’s electronic health record (EHR), which includes capture of all aspects of care, allowing patients to be prescreened according to pre-specified eligibility criteria, greatly minimising recruitment efforts.

The study incorporated standard assessments for out-of-clinic BP (24-hour ABPM and HBPM), geriatric fall-related factors (frailty, cognition, mobility, function and physical performance) and falls using a prospectively assessed falls calendar and serious fall injuries using the EHR.

Findings may not generalise to less-integrated healthcare settings or uninsured populations, younger individuals or older individuals not treated with antihypertensive medication for hypertension.

Introduction

Over 75% of US adults aged 65 years and older have hypertension, a major risk factor for cardiovascular disease (CVD).1 The CVD reduction benefits of antihypertensive medication have been demonstrated among older adults with hypertension.2 However, some studies have reported that antihypertensive medication use and intensification are associated with an increased risk of falls or serious fall injuries among older adults with hypertension.3–7 A history of falls is reported in approximately 40% of older adults taking antihypertensive medication.8 9 Approximately one-half of falls result in an injury.10 11 Serious fall injuries including fractures, brain injuries and dislocations are the most serious consequences of falls in older adults.6 Clinicians may weigh the cardiovascular benefits of antihypertensive medication intensification against the potential risk for precipitating falls in older adults with hypertension12 given falls are associated with substantial morbidity and mortality among older US. adults.11 13 14

Clinic blood pressure (BP) is most commonly used in decision-making regarding hypertension treatment intensification.15 BP measurements obtained outside of the clinic setting are often lower than measurements obtained in the clinic and can have large fluctuations throughout the day, particularly among older adults taking antihypertensive medication.16–20 Out-of-clinic BP may be a more clinically relevant physiologic measure linked to falls than clinic BP. Ambulatory BP monitoring (ABPM) measures out-of-clinic BP throughout the day and night, providing valuable diagnostic information beyond what is routinely provided in clinic.21 22 The difference between mean clinic BP and BP on ABPM is known as the white coat effect.21 ABPM is also used to assess white coat hypertension, defined as high mean BP in the clinic but not when measured outside of the clinic. Further, ABPM can assess the variability of BP over 24 hours, and the episodic decline in BP with meals (ie, postprandial BP decline).17 23 Among older adults taking antihypertensive medication, a large white coat effect,20 higher BP variability17 24 and large postprandial BP declines25 26 are common and may be associated with an increased risk of falls. However, to our knowledge, the association between BP on ABPM and falls has not been rigorously examined in a large prospective study of older adults with hypertension taking antihypertensive medication.

Home BP monitoring (HBPM) is an inexpensive and highly feasible approach to measuring BP outside of the clinic setting and has the potential to optimise the management of hypertension and reduce cardiovascular risk.27 Compared with ABPM, HBPM is more widely available in the USA and better tolerated by patients.28 Whether HBPM can identify patients for whom antihypertensive medication can be intensified without increasing the risk for a fall is unknown.

The AMBulatoRy blOod preSsure In older Adults (AMBROSIA) prospective cohort study was designed to determine whether BP on ABPM identifies older adults with hypertension taking antihypertensive medication who are at increased risk for falls. The AMBROSIA-HOME ancillary study was designed to determine whether BP on HBPM identifies older adults with hypertension taking antihypertensive medication who are at increased risk for falls, and whether HBPM provides similar predictive value for falls when compared with ABPM.

Cohort description

The AMBROSIA study enrolled 670 participants from May 2019 to November 2022. AMBROSIA was conducted at Kaiser Permanente Southern California (KPSC), an integrated healthcare delivery system that provides comprehensive services to over 4.8 million residents of Southern California. KPSC members are diverse and highly representative of the Southern California population.29 30 Participants were drawn from a recruitment base of >123 000 members age 65 years and older taking antihypertensive medication for hypertension from medical centres in the greater Los Angeles area (representing 46% of the 267 000 KPSC members aged 65 years and older taking antihypertensive medication for hypertension). Potentially eligible participants were identified using the KPSC electronic health record (EHR). Algorithms were created based on inclusion and exclusion criteria (online supplemental table 1). This strategy saved considerable effort by efficiently directing recruitment efforts via the EHR and not approaching people with little or no chance of being eligible. Potentially eligible participants were contacted by mail to introduce the study, 1068 people were prescreened by telephone to verify eligibility and 948 people were scheduled for further evaluation at the research clinic (. ABPM was initiated in 670 participants who completed two in-person study visits (Visits 1 and 2) conducted on consecutive days. A total of 656 participants completed the 24-hour ABPM and were followed for 1 year after Visit 2 to ascertain falls and serious fall injuries. Among the 656 participants who completed ABPM, 541 initiated HBPM, which was conducted over 1 week following Visit 2. Of note, 83 participants completed ABPM prior to the initiation of the AMBROSIA-HOME ancillary study, and the participation rate was 94%. Assuming a 40% fall rate in the referent group (eg, the three lowest quartiles of clinic systolic BP minus 24-hour ambulatory systolic BP), using an α=0.05, a sample size of 670 provides 80% power to detect a minimum HR of 1.45.

Figure 1

Flow diagram of participants in the AMBulatoRy blOod preSsure In older Adults (AMBROSIA) and AMBROSIA-HOME studies.

An Observational Study Monitoring Board (OSMB) of experts in hypertension, medicine and biostatistics reviewed and provided feedback on the protocol and study progress. The study was approved by the institutional review boards at Columbia University Irving Medical Centre, University of Alabama at Birmingham and KPSC. Each participant provided written consent.

Data collection

Eligible participants were invited to attend two in-person study visits. Demographic and clinical characteristics and geriatric assessments were collected over two consecutive study visits. Participants were asked to complete a 24-hour ABPM. HBPM was performed over 7 days in the subgroup participating in the AMBROSIA-HOME ancillary study following the second visit. Participants completed an ABPM Device Log to record self-reported sleep/awake times and mealtimes during the 24-hour ABPM period. We assessed frailty, cognition, mobility, function and physical performance. These assessments were chosen because they have been validated in older community-dwelling adults and can feasibly be obtained on a large study sample. A complete list of the measurements obtained is provided in online supplemental table 2.

BP measuresClinic BP

From the EHR, we extracted BP readings obtained as part of outpatient, non-urgent, clinical care within 12 weeks before study enrollment. At KPSC, attended clinic BP measurements are performed using an oscillometric device by medical staff who are certified in BP measurement.31

Research Grade BP

During the enrollment study visit, attended BP was measured in a seated position after an unattended 5 min rest using an Omron Model HEM-907XL, a validated oscillometric device. Three measures were obtained at 1 min intervals in the dominant arm followed by three measures taken at 1 min intervals in the non-dominant arm, which was the same arm on which the ABPM cuff was subsequently placed. For the assessment of orthostatic hypotension, three additional readings were obtained at 1 min intervals in the supine position, followed by readings taken after 1 and 3 min of standing. Orthostatic hypotension was defined as a decline of ≥20 mm Hg in systolic BP or ≥10 mm Hg in diastolic BP after either one or 3 min of standing. Participants with mean research grade seated clinic BP ≥180/110 mm Hg or <110/50 mm Hg were referred to their healthcare provider for evaluation and excluded from the study.

ABPM

Participants were fitted with a SpaceLabs 90 227 ABPM device, which has been previously validated. An appropriately sized cuff was placed on the participant’s non-dominant arm. Participants were instructed to wear the monitor for 24 hours and to hold their arm still when a reading was being taken. Participants were trained to complete the ABPM Device Log, which documented sleep and awake times (including naps), times when meals were started and finished and the times that the participant removed and put the cuff back on. Diaries that assess these measures are a common adjunct to ABPM. Participants were asked the time they anticipated going to sleep that night and the time they anticipated waking the next morning. The monitor was programmed to take readings every 20 min during their anticipated awake hours and every 30 min during anticipated sleep hours. The longer interval between sleep readings was chosen to increase adherence of wearing the device during sleep.

HBPM

HBPM was the core additional activity for the AMBROSIA-HOME ancillary study and occurred in the week following AMBROSIA study Visit 2 and the completion of ABPM. Participants were given a HBPM device (Microlife WatchBPhome) with an appropriately sized cuff. Study staff instructed participants on how to appropriately place the cuff on their arm, to turn on the monitor and measure their BP. Participants demonstrated their ability to measure their BP with the HBPM before leaving study Visit 2. Briefly, each participant was asked to turn on the device and measure their BP twice. In addition to turning on the monitor, placing the BP cuff in the proper position, staff ensured participants sat erect with their feet flat on the floor, arm supported on a table or pillow, had the BP cuff at heart level and rested for 5 min prior to the first measurement being taken. The WatchBPhome device was set to automatically take two measurements 1 min apart. Participants were instructed to take 2 BP readings in the morning before breakfast and two readings in the evening, 2 hours after dinner and before sleep, for the next 7 days. Participants could return the device in-person or through the mail.

Participants and their clinicians were masked to the ABPM and HBPM results to prevent BP control treatment decisions being altered.

OutcomesFalls

Data on falls in the year following Visit 2 were collected prospectively using monthly falls calendars.32 The use of calendars has been validated and used successfully in prospective long-term (>12 months) studies of falls and is the reference standard for falls assessment.32–35 A falls calendar is superior to retrospective falls assessments that require the recall of fall events over longer periods of time (ie, over the past year), which may be limited due to memory loss and susceptible to recall bias of falls with a serious injury.34 Participants were provided with 13 calendars, prepaid postage envelopes and instructions for completing and returning the falls calendars at the end of each month. Participants were asked to report ‘any fall where part of your body hits a lower surface, including falls that occur on stairs’ by marking each day of the calendar indicating if a fall had occurred or not.36 This broad definition of falls was chosen to limit variability in interpretation by participants. This definition does not require participants to make judgements about the mechanism of the fall or if the fall was accidental. Participants were asked to provide information on the fall timing, location, and if they required medical care because of the fall. Participants who did not return a fall calendar within 2 weeks after the end of the month received a telephone, email or text reminder.

Serious fall injuries

Serious fall injuries were extracted from the EHR during the prescreening assessment to assess history of serious fall injuries in the prior year and during the 1 year follow-up, based on outpatient including urgent care, emergency department and inpatient encounters with a fall-related ICD-10-CM code and an injury code for nonpathological skull, facial, cervical, claviular, humeral, forearm, pelvic, hip, fibula, tibial or ankle fractures; brain injury; or dislocation of the hip, knee, shoulder or jaw. In the absence of a fall-related code, the event was considered a fall-related injury if there was an emergency department or inpatient code for any of the serious injuries and there was no motor vehicle accident code.6

Findings to date

Preliminary findings among 250 participants related to falls during the COVID-19 pandemic were previously reported.36 Among the 656 participants of the AMBROSIA study cohort, 536 also completed HBPM. The mean (SD) number of daytime and nighttime ABPM readings was 42.1 (8.1) and 14.8 (3.9), respectively. Among participants completing HBPM, the mean (SD) number of readings per week (maximum of 28) was 25.4 (3.5), the mean (SD) number of days with two morning readings and two evening readings was 6.0 (1.5) days and 515 participants (96.1%) had three or more days with at least two morning readings and at least two evening readings. The demographic and clinical characteristics of the participants are presented in table 1. The mean age of the AMBROSIA cohort was 74.5 years, 16.5% were over 80 years of age and 55.9% were female. There were 13.1% non-Hispanic Asian or Pacific Islander, 22.3% non-Hispanic Black, 18.4% Hispanic and 44.4% non-Hispanic White participants. Over 75% of the AMBROSIA participants were overweight or obese, 64.0% had filled two or more classes of antihypertensive medication (table 1 and online supplemental figure 1), 23.3% had filled four or more other chronic medication classes, 23.6% had eGFR of <60 mL/min/1.73 m2 and 41.9% had a Charlson Comorbidity score of 3 or more. Arthritis was the most common comorbid condition (45.7%), followed by diabetes (30.2%), depression (18.8%) and coronary heart disease (16.9%). Orthostatic hypotension assessed at Visit 1 was present in 17.4% of participants. Characteristics of the subset of participants in AMBROSIA-HOME were similar. Over the 1 year follow-up, 87.5% of participants who completed ABPM returned all their calendars (median and IQR: 13; 11–13).

Table 1

Baseline characteristics of the AMBROSIA and AMBROSIA-HOME cohorts

Table 2 presents the mean (SD) systolic and diastolic BP from the EHR, research clinic, ABPM and HBPM. A summary of the distributions of the self-administered surveys is presented in table 3. Nearly 40% of the AMBROSIA participants had smoked cigarettes in their lifetime, 61.7% consumed alcohol in the past year and 39.7% reported a history of falls in the past year. Nearly one-third (30.7%) reported daily pain in the last 4 weeks and 32.6% reported dizziness or balance problems in the last 4 weeks. Nocturia was present in 92.9% of participants, 16.4% wore hearing aids, 27.9% reported their vision to be fair or poor, 17.3% used assistive devices, 19.0% reported depressive symptoms, 31.1% reported at least moderate stress, 12.6% had moderate anxiety, 5.9% had high anxiety scores and 12.3% had functional impairment. Findings from the geriatric assessments are presented in table 4. Overall, 71.8% had mild cognitive impairment, 18.7% had life-space mobility scores 60 or lower indicating lower mobility, 50.3% were pre-frail and 4.4% were frail. The use of over-the-counter medications, vitamins and supplements are presented in online supplemental table 3. Nearly 20% reported using ibuprofen in the past year, 19.1% used sleep aids, 49.1% used multivitamins, 64.8% used vitamin D and 32.7% used fish oil or Omega-3.

Table 2

Blood pressure variables in AMBROSIA and AMBROSIA-HOME

Table 3

Summary of self-administered survey measures

Table 4

Summary scores of geriatric assessments

Strengths and limitations

A primary strength of the AMBROSIA study is the racially and ethnically diverse sample of adults aged 65 years and older recruited from KPSC, a real-world clinical environment, which began implementing a hypertension programme in 2004 that has resulted in hypertension control rates exceeding 80%.31 37 The KPSC comprehensive EHR captures all aspects of medical care and a pharmacy system that captures medication orders, changes and dispenses. Additionally, the study incorporated standard assessments of out-of-clinic BP (24-hour ABPM and HBPM), geriatric fall-related factors (frailty, impaired cognition, compromised mobility, functional impairment and physical performance) and falls (a prospectively assessed falls calendar). Further, AMBROSIA examined postprandial BP decline in the participant’s usual environment. Many previous studies that assessed postprandial BP decline and postprandial hypotension have done so by measuring BP before and after a standardised meal in the clinic setting, which is not a feasible approach in routine clinical practice. Also, many prior studies rely solely on ‘research-grade’ clinic BP (ie, clinic BP which is measured under highly standardised conditions) to define on-treatment BP level and BP phenotypes including white coat effect. AMBROSIA obtained clinic BP from the EHR and measured ‘research grade’ clinic BP and can determine if the prevalence of the white coat effect differs by these two approaches. Finally, agreement between mean BP on ABPM and HBPM and the degree to which these measurement approaches are interchangeable is not known, and it is unknown whether the association of BP on HBPM with the risk for falls differs from the associations of BP measured in the clinic setting or using ABPM with falls risk. Answering these questions will inform whether HBPM can be used instead of ABPM to guide antihypertensive treatment.

Given the integrated structure of healthcare at KPSC including standardised procedures for measuring BP and managing hypertension, the results may not be generalisable to patients in less-integrated settings or uninsured populations as well as to younger individuals or older individuals not treated with antihypertensive medication for hypertension. Finally, while the finding that 71.8% of AMBROSIA participants had mild cognitive impairment assessed by the MoCA is similar with the 72% (95% CI 69% to 74%) identified in a nationally representative sample of home-dwelling US adults aged 62–90 years,38 caution is warranted when interpreting MoCA scores using the proposed threshold (<26) for identifying cognitive impairment as more research is needed to link the distribution of MoCA scores in the general population.39

Data availability statement

Data are available upon reasonable request. Anonymised data may be made available from the corresponding author on reasonable request from qualified researchers with documented evidence of training for human subjects protections and upon approval of analysis proposals with signed data-access agreements in place.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants. The study was approved by the institutional review boards at KPSC (#10699), Columbia University (#AAAS3843) and University of Alabama at Birmingham (#300003708). Participants gave informed consent to participate in the study before taking part.