Scolaris Content Display Scolaris Content Display

Base de datos Cochrane de Revisiones Sistemáticas Revisión - Intervención

Suspensión de fármacos antihipertensivos en personas mayores

Declaraciones de intereses de los autores

Versión publicada: 10 junio 2020 Historial de versiones

https://doi.org/10.1002/14651858.CD012572.pub2
Contraer todo Desplegar todo

Antecedentes

La hipertensión es un factor de riesgo importante de sufrir posteriores episodios cardiovasculares, como el accidente cerebrovascular isquémico y hemorrágico, el infarto de miocardio, la insuficiencia cardíaca, la nefropatía crónica, el deterioro cognitivo y la muerte prematura. En general, el uso de un tratamiento antihipertensivo ha reducido el riesgo de enfermedad cardiovascular y las tasas de morbilidad y de mortalidad. Sin embargo, el uso de medicamentos antihipertensivos también se asocia con efectos perjudiciales, especialmente en personas de más edad, incluida la aparición de reacciones adversas a los medicamentos e interacciones entre ellos, y puede contribuir a aumentar la carga relacionada con los medicamentos. Por ello, el retiro de los antihipertensivos puede considerarse y ser apropiado en algunas personas mayores.

Objetivos

Investigar si es factible la retirada de los fármacos antihipertensivos y evaluar los efectos de dicha retirada en la mortalidad, los resultados cardiovasculares, la hipertensión y la calidad de vida de las personas de mayor edad.

Métodos de búsqueda

El especialista en información del Grupo Cochrane de Hipertensión (Cochrane Hypertension Group) buscó ensayos controlados aleatorizados en las siguientes bases de datos hasta abril de 2019: el registro especializado del Grupo Cochrane de Hipertensión (Cochrane Hypertension Specialised Register), CENTRAL (número 3, 2019), Ovid MEDLINE, Ovid Embase, la Plataforma de registros internacionales de ensayos clínicos de la OMS y en ClinicalTrials.gov. También se verificaron las referencias, se buscaron las citas y, si fue necesario, se estableció contacto con los autores de los estudios para identificar estudios adicionales. En la búsqueda no hubo restricciones de idioma.

Criterios de selección

Se incluyeron ensayos controlados aleatorizados (ECA) de retiro versus continuación de los fármacos antihipertensivos utilizados para la hipertensión o la prevención primaria de las enfermedades cardiovasculares en adultos mayores (definidos como de 50 años o más). Los participantes se consideraron aptos si vivían en su domicilio, en centros residenciales de atención a la tercera edad o en contextos hospitalarios. Se intentó incluir ensayos que estudiaran la retirada completa de la medicación antihipertensiva, y aquellos que se centraran en la reducción de la dosis del fármaco antihipertensivo.

Obtención y análisis de los datos

Se comparó la intervención de suspender o reducir la medicación antihipertensiva con el tratamiento habitual mediante diferencias de medias (DM) e intervalos de confianza del 95% (IC del 95%) para las variables continuas y se utilizaron los odds‐ratios (OR) de Peto y el IC del 95% para las variables binarias. Los resultados primarios fueron: mortalidad, infarto de miocardio, aparición de reacciones adversas a los medicamentos o reacciones adversas a la retirada de los mismos. Los resultados secundarios incluyeron: presión arterial, hospitalización, accidente cerebrovascular, éxito de la retirada de los antihipertensivos, calidad de vida y caídas. Dos autores, de forma independiente y por duplicado, realizaron todos los estadios de la selección de estudios, la extracción de datos y la evaluación de la calidad.

Resultados principales

Seis ECA cumplieron los criterios de inclusión y fueron incluidos en la revisión (1073 participantes). La duración de los estudios varió entre 4 y 56 semanas. El metanálisis de los estudios mostró que, en el grupo de retiro, en comparación con el de continuación, las probabilidades de mortalidad por todas las causas fue de 2,08 (IC del 95%: 0,79 a 5,46; evidencia de certeza baja), para el infarto de miocardio 1,86 (IC del 95%: 0,19 a 17,98; evidencia de certeza muy baja) y para el accidente cerebrovascular 1,44 (IC del 95%: 0,25 a 8,35; evidencia de certeza baja). La presión arterial fue más alta en el grupo de retiro que en el grupo de continuación (presión arterial sistólica: DM = 9,75 mmHg, IC del 95%: 7,33 a 12,18; y presión arterial diastólica: DM = 3,5 mmHg; IC del 95%: 1,82 a 5,18; evidencia de certeza baja). En cuanto a la aparición de eventos adversos, no fue posible realizar un metanálisis; el retiro de los antihipertensivos no pareció aumentar el riesgo de eventos adversos y podría conducir a la resolución de las reacciones adversas al fármaco, aunque los estudios elegibles presentaron información limitados sobre los efectos adversos de la retirada de los fármacos (evidencia de certeza muy baja). Un estudio informó sobre la hospitalización con un odds ratio de 0,83 para la suspensión en comparación con la continuación (IC del 95%: 0,33 a 2,10; evidencia de certeza baja). No se identificaron estudios que informaran acerca de caídas. Entre el 10,5% y el 33,3% de los participantes del grupo de retiro, en comparación con el 9% y el 15% del grupo de continuación, experimentaron un aumento de la presión arterial u otros criterios clínicos (según lo predefinido por los estudios) que requerirían reanudar el tratamiento/abandonar el estudio. Las fuentes de sesgo incluyeron el informe selectivo (sesgo de notificación), la falta de cegamiento de la evaluación de los resultados (sesgo de detección), datos de resultados incompletos (sesgo de desgaste) y falta de cegamiento de los participantes y el personal (sesgo de realización).

Conclusiones de los autores

No existe evidencia de que la interrupción del tratamiento, en comparación con la continuidad de los antihipertensivos utilizados para la hipertensión o la prevención primaria de las enfermedades cardiovasculares en los adultos mayores, tenga un efecto en la mortalidad por todas las causas y el infarto de miocardio. La evidencia fue de certeza baja o muy baja, principalmente debido al pequeño tamaño de los estudios y a las bajas tasas de eventos. Estas limitaciones impiden establecer conclusiones firmes sobre el efecto de retirar la prescripción de antihipertensivos en estas variables. Las investigaciones futuras deberían centrarse en poblaciones con mayor incertidumbre en cuanto a la relación beneficio/riesgo del uso de medicación antihipertensiva, como las personas con fragilidad, los grupos de edad más avanzada y aquellos con polifarmacia, y medir resultados clínicamente importantes como las caídas, la calidad de vida y los efectos adversos de los medicamentos.

PICO

Population
Intervention
Comparison
Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Retirada de fármacos para la hipertensión a personas mayores

Objetivo

Esta revisión tuvo como objetivo averiguar si es posible suspender la administración de medicamentos para la hipertensión a personas mayores. También se quiso averiguar los efectos de retirar esta medicación.

Se incluyó a adultos de 50 años o más que tomaban medicamentos para la tensión alta (hipertensión) o para prevenir enfermedades cardíacas (prevención primaria). Se excluyeron los estudios con personas que habían sufrido previamente un ataque cardíaco, un ictus u otra enfermedad cardíaca (prevención secundaria).

Se comparó la suspensión o reducción de la dosis de fármacos para la tensión arterial con la continuación de los fármacos para la tensión arterial.

Antecedentes

La tensión alta, también conocida como hipertensión, es un factor de riesgo de muchas enfermedades, como el ataque al corazón, la insuficiencia renal y el accidente cerebrovascular. Si bien la hipertensión no suele producir síntomas, mantener la presión arterial bajo control es vital para preservar la salud y reducir el riesgo de afecciones graves.

La hipertensión suele controlarse con hábitos de vida y medicamentos para la tensión (antihipertensivos). Hay muchos tipos diferentes de medicamentos para la tensión.

Los antihipertensivos pueden causar efectos secundarios peligrosos, como mareos y fatiga que pueden provocar caídas. Las personas mayores corren un mayor riesgo de sufrir efectos secundarios de los medicamentos en comparación con las personas más jóvenes. No está claro si los efectos beneficiosos de los agentes antihipertensivos superan los efectos perjudiciales en las personas mayores.

Características de los estudios

La búsqueda hasta abril de 2019 encontró seis estudios, que incluyen a 1073 adultos mayores en total. La edad media de las personas de los estudios fue de 58 a 82 años. En tres de los estudios, la dosis del antihipertensivo se redujo lentamente antes de retirarse.

Resultados clave

Se halló que es posible suspender la administración de antihipertensivos en adultos mayores. La mayoría de las personas mayores en los grupos de retiro no necesitaron reiniciar la medicación.

Se halló evidencia de certeza baja de que suspender la medicación antihipertensiva aumentó la presión sanguínea ligeramente.

Se encontró evidencia de certeza baja o muy baja de que suspender los medicamentos para la presión arterial no aumentó el riesgo de sufrir un ataque cardíaco, un accidente cerebrovascular, hospitalización ni de morir.

Se encontró evidencia de certeza muy baja de que la suspensión de los medicamentos para la presión arterial no aumentó el riesgo de eventos adversos y podría resolver los efectos secundarios, pero esto no se informó bien y, por lo tanto, no es posible establecer conclusiones.

Ninguno de los estudios informó si retirar la medicación para la hipertensión afectó a las caídas.

Certa Calidad de las pruebas

Se calificó la certeza de la evidencia usando cuatro niveles: muy baja, baja, moderada o alta. La evidencia de certeza alta implica que hay mucha confianza en los resultados. La evidencia de certeza muy baja implica muy poca confianza en los resultados. La certeza de la evidencia se calificó como muy baja y baja.

Conclusión

Podría ser seguro retirar la medicación antihipertensiva a personas mayores que toman medicamentos para la tensión o para la prevención primaria de la enfermedad cardíaca.

Los adultos mayores no deben dejar de tomar su medicación sin consultar a un profesional de la salud.

Los estudios futuros deben incluir a adultos mayores que toman otros varios medicamentos o que presentan fragilidad.

Authors' conclusions

Implications for practice

The findings of this review have clinical implications for the management of hypertension in older adults, and could be considered in the next updates of international guidelines. For example, the 2017 American College of Cardiology/American Heart Association (ACC/AHA) guidelines (Whelton 2018) did not comment on withdrawal of antihypertensives beyond perioperative recommendations, although they do provide information on the risks of abrupt cessation of beta‐blockers and clonidine (physiological adverse drug withdrawal reactions). The 2018 European Society of Cardiology/European Society of Hypertension (ESC/ESH) guidelines (Williams 2018) stated, ‘Withdrawal of BP‐lowering drug treatment on the basis of age, even when patients attain an age of ≥ 80 years, is not recommended, provided that treatment is well tolerated’. It also commented that ‘it is well established that BP‐lowering treatment withdrawal leads to a marked increase in CV disease’, based on a subgroup analysis of the HYVET trial, which was designed to investigate treatment versus placebo rather than withdrawal of treatment. Our systematic review of deprescribing trials did not confirm this risk. Future iterations of treatment guidelines could consider also including data from deprescribing trials synthesised in our review.

The limitations of the available data on deprescribing of antihypertensives, as described in the discussion, restrict their impact on current practice. There is a paucity of high certainty evidence applicable to current practice and to the situations in which deprescribing is often considered clinically (frailty, falls, dementia, polypharmacy), and an absence of data on key outcomes important to older adults, such as falls and quality of life. The variability in the populations and inclusion/exclusion criteria of the included studies mean that we can not make recommendations as to which individuals should have their antihypertensive medication deprescribed, or a BP cut‐point at which deprescribing should occur. However, the feasibility of withdrawal and the lack of a significant increase in mortality, CVD or stroke in meta‐analysis, may help clinicians when considering deprescribing antihypertensives and discussing this with patients, particularly in those who are at high risk of harm, such as those with frailty, falls, dementia and polypharmacy. Shared decision‐making, specification of patient‐specific goals (such as reversal of side effects or reduction of pill burden) and close monitoring of the effects in the individual remain pillars of clinical practice when deprescribing antihypertensives (Scott 2019).

Implications for research

The current state of evidence on antihypertensive discontinuation reveals several areas of focus for future research. More RCTs will help to further establish the potential benefit and safety of antihypertensive withdrawal, particularly in populations where there is uncertainty in the net benefit of treatment. Future RCTs would ideally be powered to detect differences in clinically meaningful outcomes such as cardiovascular events or mortality. These RCTs could also focus on evaluating the comparative effects of discontinuing different medication classes. For example, investigating whether there are differences in clinical outcomes, rates of withdrawal symptoms, and success rate between different medications or medication classes. Since the context for discontinuation may be different in certain populations (e.g. frail older person with multiple comorbidities versus low‐risk younger old), it will also be helpful for future studies to examine whether differences in the potential benefit and safety of antihypertensive discontinuation exist between specific populations. It will also be useful for future studies to clearly describe their discontinuation plan (e.g. tapering rate and duration), such that discontinuation protocols can be replicated in other studies and/or more easily implemented in practice. Since risk of future cardiovascular events is of particular interest with respect to antihypertensive discontinuation, it will also be helpful for future studies to gather longer‐term outcome data (12 months or beyond). This would serve to more clearly elucidate the downstream effects of antihypertensive discontinuation on cardiovascular events or mortality. Continuing to follow‐up participants who have restarted the medication due to predefined BP or clinical criteria would provide important information for shared decision‐making. Further, since return of underlying symptoms or adverse drug withdrawal events are also particularly relevant for antihypertensive discontinuation, future RCTs should include robust measures to capture this to provide a better picture of the safety of discontinuation. Finally, future studies should focus on evaluating patient‐centred outcomes such as quality of life and falls.

In broader terms, this review highlights the need for more research into the benefits and harms of deprescribing specific medication classes to provide evidence to guide clinical decision‐making and inform clinical practice guidelines. It will also be important to consider whether large RCTs are the best study design for this research question and the role of large observational database studies to inform decision‐making. Additionally, as identified in ours and previous reviews (Froom 1997; Iyer 2008), identifying deprescribing studies for systematic reviews can be challenging due to the wide variety of terminology in this space. Development of a standardised search strategy and recommendations for keyword referencing for all new studies in the field is likely to improve the ability to identify and synthesise evidence about deprescribing. The need for further research and synthesis of deprescribing research aligns with the internationally recognised problem of medical excess, such as overdiagnosis, overtreatment, and unnecessary testing (Johansson 2019).

Summary of findings

Open in table viewer
Summary of findings 1. Discontinuation by no treatment/placebo of antihypertensives compared to continuation in older people

Discontinuation by no treatment/placebo of antihypertensives compared to continuation in older people

Patient or population: Older adults, 50 years and older
Setting: All settings
Intervention: Discontinuation by no treatment/placebo of antihypertensives
Comparison: Continuation

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with Continuation

Risk with Discontinuation by no treatment/placebo of antihypertensives

All‐cause mortality
follow‐up: range 12 weeks to 12 months

Study population

OR 2.08
(0.79 to 5.46)

630
(4 RCTs)

⊕⊕⊝⊝
LOW 1 2 3 4

19 per 1,000

40 per 1,000
(15 to 98)

Moderate

26 per 1,000

52 per 1,000
(20 to 126)

Myocardial infarction (fatal and nonfatal)
follow‐up: range 16 weeks to 12 months

Study population

OR 1.86
(0.19 to 17.98)

447
(2 RCTs)

⊕⊝⊝⊝
VERY LOW 5 6

5 per 1,000

9 per 1,000
(1 to 77)

Moderate

3 per 1,000

5 per 1,000
(1 to 46)

Hospitalisation
follow‐up: 16 weeks

Study population

OR 0.83
(0.33 to 2.10)

385
(1 RCT)

⊕⊕⊝⊝
LOW 7

54 per 1,000

45 per 1,000
(18 to 107)

Stroke (fatal + nonfatal + TIA)
follow‐up: range 16 weeks to 12 months

Study population

OR 1.44
(0.25 to 8.35)

524
(3 RCTs)

⊕⊕⊝⊝
LOW 38

8 per 1,000

11 per 1,000
(2 to 62)

Moderate

5 per 1,000

8 per 1,000
(1 to 43)

Systolic blood pressure
follow‐up: range 12 weeks to 12 months

The mean systolic blood pressure ranged from 123 to 145 mmHg 9

MD 9.75 mmHg higher
(7.33 higher to 12.18 higher)

‐

767
(5 RCTs)

⊕⊕⊝⊝
LOW 10 11 12

Diastolic blood pressure: range 12 weeks to 12 months

The mean diastolic blood pressure ranged from 70‐95 mmHg 9

MD 3.5 mmHg higher
(1.82 higher to 5.18 higher)

‐

768
(5 RCTs)

⊕⊕⊝⊝
LOW 10 12 13

Adverse drug reactions and adverse drug withdrawal reactions (adverse reactions)
follow‐up: range 12 weeks to 12 months

Overall there was some reversal of adverse drug reactions in the discontinuation group, otherwise no change reported.

One study reported no difference in frequency of side effects although data were not shown.

One study reported reversal of slight postural drop and improvement in renal function and serum cholesterol in the discontinuation but not the continuation group.

One study reported that more participants experienced ankle oedema in the discontinuation than the continuation group, however, statistical significance was not reported.

All three studies reported changes in potassium in the discontinuation group.

‐

245
(3 RCTs)

⊕⊝⊝⊝
VERY LOW 14 15

Falls ‐ not reported

None of the included studies reported on falls or falls risk.

‐

‐

‐

*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;

GRADE Working Group grades of evidence
High certainty: We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect
Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

1 Not downgraded for risk of bias: some concern about attrition bias in 2 of the studies (high and uneven drop out rates), not enough to assess as serious but considered when downgrading in other areas.

2 Not downgraded for inconsistency but considered when downgrading other areas: some concern about heterogeneity due to the difference in mortality rates between Myers 1982 and Maland 1983 but likely explained by differences in populations (Maland's participants were older and recruited from geriatric institutions).

3 Downgraded one level for imprecision, the total number of events was very low ; additionally, the CI included the null effect and appreciable benefit favouring continuation.

4 Downgraded one level for indirectness due to concern about the age of the majority of studies relevant for this outcome ‐ standards of care and recommendations for treatment have significantly changed over the past 35 years. Downgrading in this category also took into account the large concern about imprecision (which was downgraded 1, but had considered downgrading 2 steps) as well as potential risk of bias and inconsistency.

5 Downgraded one level for risk of bias as neither study had blinding of participants or physicians; possible that diagnosis of MI could have been influenced by knowledge of intervention.

6 Downgraded two levels for imprecision as the total number of events was very low; additionally, the CI included the null effect and appreciable benefit favouring continuation.

7 Downgraded two levels for imprecision as the total number of events was very low; additionally, the CI included the null effect and appreciable benefit favouring discontinuation.

8 Downgraded one level for indrectness due to concern about the age of the majority of studies relevant for this outcome ‐ standards of care and recommendations for treatment have significantly changed over the past 35 years. Downgrading in this category also took into account the large concern about imprecision (which was downgraded 1, but had considered downgrading 2 steps)..

9 Range of BP at end of follow‐up period in continuation group of included studies.

10 Downgraded one level for risk of bias due to concern about lack of blinding of participants and physicians in two of the studies, uneven dropouts in two of the studies and reporting bias in one of the studies.

11 Downgraded one level for inconsistency due to substantial heterogeneity, subgroup analyses based on duration of follow‐up and class of medication was not able to explain heterogeneity, and no other cause identified.

12 Some concern about indirectness as BP is a surrogate marker, also the majority of studies are more than 35 years old. Did not downgrade in this category, however contributed to the decision to downgrade in the 'Risk of bias' category.

13 Downgraded one level for inconsistency; duration of follow‐up was able to explain some of the inconsistency, however, heterogeneity remained in the subgroup with a duration of follow‐up 12 months or greater.

14 Downgraded two levels for risk of bias due to very serious concern about lack of blinding, attrition bias and poor detection and reporting of outcome.

15 Downgraded one level for imprecision due to small sample size and number of events.

TIA: Transient ischaemic attack

Background

Description of the condition

Hypertension or high blood pressure (BP) increases with age, and is prevalent in over two‐thirds of people aged over 60 (Fryar 2017). Hypertension is a major risk factor for subsequent cardiovascular events including ischaemic and haemorrhagic stroke, myocardial infarction, heart failure, chronic kidney disease, cognitive decline and premature death (NICE 2019). In older adults, the use of antihypertensive medications has led to reductions in overall cardiovascular disease (CVD), morbidity rates and mortality rates in people with high BP (Ikeda 2014; Musini 2019).

Ensuring appropriate use of antihypertensive drugs can be challenging, especially in an older population with increasing age‐associated pathologies, including polypharmacy, multi‐morbidity, frailty, orthostatic hypotension (a significant decrease in blood pressure when changing from a sitting or lying position to standing), falls and cognitive impairment (Parekh 2017). Antihypertensive medications can increase the risk of adverse drug reactions (ADRs), and cause undesired metabolic effects such as hypokalaemia (low potassium in the blood), hyperkalaemia (high potassium in the blood), hyperglycaemia (high blood sugar level) or hyperuricaemia (excess of uric acid in the blood). Blood pressure lowering in older people may decrease cerebral autoregulation resulting in worsening of cognition, and a higher BP may be preferred to ensure adequate cerebral blood flow (Goshgarian 2019). Antihypertensives may contribute to polypharmacy, which is the use of multiple medicines simultaneously in an individual patient, and has been widely documented as a risk factor for ADRs, drug‐disease and drug‐drug interactions as well as increased morbidity and mortality, and costs (Reeve 2014; Steinman 2006; Wastesson 2018). Reducing the number of medications taken including antihypertensive medications by deprescribing (i.e. planned and supervised withdrawal of medications that are inappropriate), may therefore lead to reduced adverse effects and improved quality of life (QoL) in older people (Gnjidic 2014; Reeve 2014; Scott 2015).

Description of the intervention

Hypertension is treated with lifestyle and medications. Healthy lifestyle measures include a diet low in salt, regular exercise, weight loss, safe alcohol consumption, avoidance of excessive caffeine consumption, and smoking cessation. Several antihypertensive classes can be used either alone or in combination, including angiotensin receptor blockers, angiotensin converting enzyme inhibitors, calcium channel blockers, and diuretics, most commonly thiazide diuretics. Other medication classes can be added if hypertension is resistant (National Heart Foundation of Australia 2016; NICE 2019; Whelton 2018; Williams 2018). However, over time, the benefits and harms of a medication in an individual can change, making medications that were once appropriate, now inappropriate (that is, the likely harms outweigh the likely benefits in the individual, including where the benefits no longer align with the person's goals of care) (Reeve 2015; Reeve 2017; Scott 2015).

Withdrawal of antihypertensive medications can either be complete (immediate) discontinuation of medications or dose reduction with or without intermittent therapy reduction strategies, also known as tapered withdrawal (gradual withdrawal according to a predefined dosing schedule or following clinical response) (Ekbom 1994: Reeve 2014). In the context of this review, we included and evaluated randomised controlled trials (RCTs) that withdrew antihypertensive medications in older adults, either by immediate discontinuation or by tapering interventions.

How the intervention might work

Withdrawal of antihypertensive drugs in older people, prescribed one or more antihypertensive medications for hypertension or primary prevention of cardiovascular disease, may theoretically cause a reduction in undesired metabolic effects and reduce medication errors, drug‐drug and drug‐disease interactions, and ADRs (that may occur as a result of continued use of antihypertensive medications). Additional possible positive effects of antihypertensive drug withdrawal may include a reduction in the risk of falls, reduction in compromised cerebral blood flow and hypoperfusion (reduced blood flow) (Froom 1997; Goshgarian 2019; Scott 2019). However, withdrawal of antihypertensive medications may also cause an increase in BP and may increase the risk of cardiovascular outcomes or mortality (Ekbom 1994; Ikeda 2014; NICE 2019).

Why it is important to do this review

There is substantial evidence that the use of antihypertensive medications within the context of polypharmacy and multi‐morbidity can lead to increased risk of harm in older people (Scott 2019; Woolcott 2009). To inform the appropriateness of withdrawing antihypertensive medications in older people, we thus proposed to critically evaluate the evidence in relation to the safety and efficacy of withdrawal of antihypertensive medications to inform clinical decisions and future research. Previous systematic reviews and meta‐analyses included multiple medication classes without a primary focus on antihypertensive withdrawal interventions (Iyer 2008), were conducted in the 1990s (Froom 1997), focused on all ages, not specifically on older people (Nelson 2001; Van der Wardt 2017), or primarily focused on the effect of medication withdrawal interventions on cognition in older people (Jongstra 2016). Thus, this review provides up‐to‐date evidence and investigates a number of clinically relevant outcomes of antihypertensive withdrawal in older people.

Objectives

To investigate whether withdrawal of antihypertensive medications is feasible, and evaluate the effects of withdrawal of antihypertensive medications on mortality, cardiovascular outcomes, hypertension and quality of life in older people.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised controlled trials. All other study types were excluded e.g. observational studies, case series.

Types of participants

Participants were adults aged 50 years and over prescribed one or more antihypertensive medication(s) for hypertension or primary prevention of cardiovascular disease living in the community, residential aged care facilities or in hospital settings. The cut‐off of 50 years to define ‘older adults’ was chosen to maximise inclusion of relevant older studies and studies from developing countries (Shenkin 2017).

To be eligible, either all participants had to be aged 50 years or older, or results for participants aged 50 years and older had to be presented in a separate subgroup analysis, or the majority of participants had to be 50 years or older (as determined by looking at mean/median age and standard deviation (SD)/interquartile range (IQR) (to be included, the mean age minus the SD must be ≥ 50 years, or if IQR, three‐quarters of participants must be ≥ 50 years old, or the article reported the number of people ≥ 50 years old).

Studies were included if the reported indication in the study was hypertension or primary prevention of cardiovascular disease and less than 20% of the population had cardiovascular disease at baseline.

Types of interventions

Included studies assessed withdrawal of antihypertensive medications in older adults prescribed for hypertension or primary prevention of cardiovascular disease. Withdrawal of medications may be through abrupt withdrawal, tapering to complete withdrawal or dose reduction. The control intervention included no withdrawal of antihypertensive medications (i.e. continuation).

The following antihypertensive medications were included:

  • diuretics which act primarily by blocking reabsorption of sodium at four major sites in the nephron. Different classes of diuretics act at different sites. Loop diuretics (e.g. furosemide, torsemide) act in the thick ascending limb of the loop of Henle. Thiazide‐type diuretics (e.g. hydrochlorothiazide, chlorthalidone, indapamide) act in the distal tubule and connecting segment. Potassium‐sparing diuretics (e.g. amiloride, triamterene) increase diuresis, but without causing potassium to be lost from the body. Aldosterone receptor antagonists (e.g. spironolactone, eplerenone) stop the entry of aldosterone into the principal cells of the collecting duct and late distal tubule of the nephron, which prevents sodium and water retention;

  • beta‐blockers (e.g. atenolol, carvedilol) block the effects of catecholamines at receptor sites in the heart, peripheral vasculature, bronchi, pancreas, uterus, kidney, brain and liver. Beta‐blockers reduce BP by blocking the effects of catecholamine on the heart and blood vessels;

  • ACE inhibitors (e.g. captopril, enalapril) act by blocking the renin‐angiotensin system; specifically, they block conversion of angiotensin I to angiotensin II and bradykinin. ACE inhibitors reduce the effects of angiotensin II‐induced vasoconstriction, sodium retention and aldosterone release;

  • calcium channel blockers (e.g. amlodipine, felodipine) act by blocking inward current of calcium via L‐type calcium channels. Calcium channel blockers lower BP by blocking the effects of calcium on blood vessels;

  • angiotensin II receptor antagonists (e.g. candesartan, irbesartan) act by blocking binding of angiotensin II to type 1 angiotensin (AT1) receptors. This leads to reduction in angiotensin II‐induced vasoconstriction, sodium reabsorption and aldosterone release; blood vessels dilate leading to reduction in BP;

  • renin inhibitors (e.g. aliskiren) prevent the conversion of angiotensinogen to angiotensin I by binding to the S3bp binding site of renin.

Types of outcome measures

Primary outcomes

  • Mortality (all‐cause mortality, cardiovascular mortality).

  • Myocardial infarction (fatal and non‐fatal).

  • Adverse drug reactions and adverse drug withdrawal reactions.

Secondary outcomes

  • Blood pressure (BP), including systolic and diastolic BP, before and after withdrawal of antihypertensive drugs and mean arterial pressure.

  • Hospitalisation (all‐cause, cardiovascular hospitalisation, heart failure hospitalisation).

  • Stroke (fatal and non‐fatal, ischaemic and haemorrhagic, transient ischaemic attack).

  • Success (rate) of withdrawal from antihypertensive drugs over the short term (12 months or less) and long term (greater than 12 months). Success (rate) will be defined as the ability of the participant to complete the study having experienced withdrawal from antihypertensive medications and resisted restarting existing treatment given before withdrawal.

  • Quality of life (QoL) of participants, carers, families or a combination, measured with validated QoL instruments (e.g. EuroQol ‐ five dimensions questionnaire (EQ‐5D), Short Form ‐ six dimensions (SF‐6D)).

  • Falls

There was no restriction on duration of follow‐up for any of the outcomes.

Search methods for identification of studies

Electronic searches

The Cochrane Hypertension Information Specialist searched the following databases without language, publication year or publication status restrictions:

  • the Cochrane Hypertension Specialised Register via the Cochrane Register of Studies (CRS‐Web) (searched 22 May 2019);

  • the Cochrane Central Register of Controlled Trials (CENTRAL, 2019, Issue 3) via the Cochrane Register of Studies (CRS‐Web) (searched 22 April 2019);

  • MEDLINE Ovid (from 1946 onwards), MEDLINE Ovid Epub Ahead of Print, and MEDLINE Ovid In‐Process & Other Non‐Indexed Citations (searched 22 April 2019);

  • Embase Ovid (from 1974 onwards) (searched 22 April 2019);

  • ClinicalTrials.gov (www.clinicaltrials.gov) (searched 22 April 2019);

  • World Health Organisation International Clinical Trials Registry Platform (http:// https://apps.who.int/trialsearch) (searched 22 April 2019).

The Information Specialist modelled subject strategies for databases on the search strategy designed for MEDLINE. Where appropriate, they were combined with subject strategy adaptations of the highly sensitivity and precision‐maximising search strategy designed by Cochrane for identifying randomised controlled trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0, Box 6.4.d. (Higgins 2011)). We present search strategies for major databases in Appendix 1.

Searching other resources

The Cochrane Hypertension Information Specialist searched the Hypertension Specialised Register segment (which includes searches of MEDLINE and Embase) for systematic reviews and Epistemonikos to retrieve published systematic reviews related to this review title to identify additional relevant trials. The Cochrane Hypertension Information Specialist searched the Hypertension Specialised Register segment for information of adverse effects relevant to this review.

We checked the bibliographies of included studies and any relevant systematic reviews identified for further references to relevant trials.

We contacted experts/organisations in the field to obtain additional information on relevant trials.

We contacted trial authors for clarification and further data if trial reports were unclear.

Data collection and analysis

Selection of studies

Two review authors (out of ER, WT, MS, AT, TM, IH, VJ, and DG) independently conducted article screening for relevance and adherence to inclusion criteria. If studies did not meet the inclusion criteria, we excluded them and recorded reasons for exclusion.  Disagreements were resolved through consultation with a third review author.

Data extraction and management

Two review authors (out of ER, WT, MS, AT, and DG) independently performed data extraction. Disagreements were resolved by consultation with a third review author.

The summary statistics required for each trial and outcome for continuous data included the values at different time points, mean change from baseline or difference between intervention and control group, the standard deviation and the number of participants in each group (discontinuation and continuation). The baseline assessments were defined as the latest available assessment between the discontinuation and continuation group, or from baseline. For binary outcomes (e.g. success rate), the number and percentage in (each) group were sought.

We extracted the following data from the studies and presented them in a summary table:

  • author, year of publication, country;

  • type of intervention;

  • antihypertensive medication withdrawn;

  • withdrawal method (immediate or tapered);

For each outcome measure, we extracted data as per the primary analysis presented in each of the studies.

Assessment of risk of bias in included studies

Two review authors (out of ER, WT, MS, AD, and DG) independently assessed the risk of bias in the included studies using the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011; Sterne 2014). We assessed the risk of bias in terms of internal validity criteria for RCTs including random sequence generation (randomisation), allocation concealment, participant and study personnel blinding, blinding of outcome assessors, level of incomplete outcome data, selective reporting and other risks of bias that may be relevant (Higgins 2011; Sterne 2014). Disagreements were resolved through consultation with a third review author, when necessary.

Measures of treatment effect

Studies may or may not utilise similar rating scales in outcome assessment. For this reason, for continuous outcomes, we used the mean difference (MD) when the collective studies utilised identical scales of rating or tests. We planned to use the standardised mean difference (SMD) if dissimilar scales of rates or tests were used. In the case of binary outcomes such as mortality, we used a Peto odds ratios (ORs) to measure the treatment effect. We also calculated 95% confidence intervals (Cls).

Unit of analysis issues

If there were any cluster RCTs, we planned to determine if the risk of unit of analysis error was dealt with appropriately. Where the analysis was carried out correctly taking into account the clustering design, we planned to consider the studies for meta‐analysis and use the reported effect sizes and standard errors. Where the analysis was incorrect (i.e. not taking the clustering design into account), we planned to apply an interclass correlation (Higgins 2011). However, no cluster RCTs were found.

Dealing with missing data

We contacted the corresponding author of included studies in the event of missing data that would compromise the ability of the review authors to examine the data and eligibility for study exclusion/inclusion in the final analysis.

Assessment of heterogeneity

We performed meta‐analysis where studies were satisfactorily homogenous in terms of interventions, outcomes and participants. In the evaluation of heterogeneity, we determined clinical heterogeneity by review author opinion and used an I2 test to determine statistical heterogeneity. An I2 value higher than 50% was considered as evidence for the presence of substantial heterogeneity of the studies. If substantial heterogeneity was present, we aimed to investigate the reasons for the presence of heterogeneity through subgroup analysis.

Assessment of reporting biases

Two review authors (out of ER, DG, MS, AT, and WT) independently assessed the risk of reporting bias in the included studies following the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We resolved disagreements through consultation with a third review author.

Data synthesis

We synthesised data for each study outcome separately using Review Manager 5 (RevMan 2014), and we undertook meta‐analyses for outcomes reported in multiple studies.

We compared outcome measures for binary data utilising Peto odds ratios, due to the rare nature of the primary outcomes, and 95% CI. This necessitated the use of the fixed effect model. For continuous variables or data, we utilised mean differences and 95% CI.

Subgroup analysis and investigation of heterogeneity

Where there were two or more studies, we performed subgroup analysis of medications within antihypertensive medication classes (e.g. ACE inhibitors and diuretics) and based on duration of follow‐up (less than 12 months versus 12 months or longer) if sufficient and meaningful data were available. We used class of medication and duration of follow‐up subgroups when investigating heterogeneity. Additionally, we intended to conduct subgroups analyses for age group (50 to 65 years, over 65 to 80 years, over 80 years) and gender (men and women). However, this was not possible as data were not reported separately for these groups.

Sensitivity analysis

We intended to perform sensitivity analysis to consider how the results of any meta‐analyses undertaken changed under different assumptions, related to the reasons for these effects. More specifically, we intended to conduct a sensitivity analysis on the choice of utilising a random‐effects model (Higgins 2011). However, we were unable to perform any sensitivity analyses due to the limited data and small number of studies identified.

Summary of findings and assessment of the certainty of the evidence

We used the GRADE approach to assess the certainty of the evidence for each outcome (Schünemann 2011a; Schünemann 2011b). We present key findings of the review, including a summary of the data, the magnitude of the effect size and the overall certainty of the evidence, in summary of findings Table 1. We preselected the following outcomes for inclusion in the summary of findings Table 1: mortality, myocardial infarction, adverse drug reactions, blood pressure (systolic and diastolic), hospitalisation, stroke and incidence of falls.

Results

Description of studies

Studies are described in detail in Characteristics of included studies; Characteristics of excluded studies and summarised in Table 1.

Open in table viewer
Table 1. Summary of studies

Study, country

Setting

Study duration

Antihypertensive medication class withdrawn

Discontinuation plan

Burr 1977, Wales

Long‐stay geriatric wards

12 months

Diuretic

Not specified

Langford 1984, USA

Primary care

56 weeks

Any antihypertensive medication

Participants were withdrawn from therapy using a standardised step‐down withdrawal programme taking from 2 to 8 weeks, depending on the number and dosage of drugs at entry. The diuretic was the last agent withdrawn.

Drug withdrawal took place in a stepped fashion with the highest step drug being removed first. Target time for withdrawal was 6 weeks.

Maland 1983, USA

Community

12 months

Diuretic

Not specified

Moonen 2015, Netherlands

128 general practices

16 weeks

β‐Blocker, diuretic, angiotensin‐converting enzyme inhibitor, angiotensin receptor blocker or calcium channel blocker.

The discontinuation of antihypertensive treatment was performed by the participant’s physician according to an algorithm composed by the investigators.

All physicians were instructed to withdraw antihypertensive treatment until a maximum increase of 20 mmHg in SBP was reached.

The physician monitored BP every week until no further changes in antihypertensive treatment were made

Myers 1982, Canada

Geriatric institution

12 months

Diuretic

Not specified

Walma 1997, Netherlands

8 general practices

6 months

Diuretic

Participants with frusemide dosages of 40 or 80 mg/day went through a dose‐halving regimen of one and two weeks, respectively, to prevent severe rebound effects.

Dose halving was started immediately after randomisation and was performed double‐blind.

BP: Blood pressure
SBP: Systolic blood pressure

Results of the search

A total of 3677 articles were identified. After de‐duplication, title and abstract screening, a total of 39 studies were obtained in full text (Figure 1). We excluded 33 articles (Characteristics of excluded studies). No additional eligible studies were found through hand searching of reference lists.

Figure 1
Study flow diagram.

Study flow diagram.

Included studies

Six studies fulfilled the inclusion criteria, comprising 1073 participants (Characteristics of included studies). Two studies were conducted in the USA (Langford 1984; Maland 1983), two in the Netherlands (Moonen 2015; Walma 1997), one in Canada (Myers 1982) and one in Wales (Burr 1977).

Design

The six studies were parallel‐group RCTs. There was a wide range of study duration and follow‐up, ranging from four weeks to 56 weeks.

Sample Size

All studies included a relatively small number of participants. Two studies included fewer than 100 participants, and four studies included between 100 and 400 participants.

Study setting and participants

Four studies included participants in general practice or primary care (Langford 1984; Maland 1983; Moonen 2015; Walma 1997), one study included participants in long‐stay geriatric wards (Burr 1977), and one study included participants living in a geriatric institution (Myers 1982). Mean age ranged from 57.5 years (Langford 1984) to 82.0 years (Burr 1977). Two studies (Moonen 2015; Walma 1997) did not report overall mean age. With respect to inclusion criteria, overall, trials included different study populations. The study by Burr 1977 included participants using diuretics for over one month with no history of significant cardiovascular events (e.g. congestive cardiac failure). Maland 1983 included participants with an average DBP of 90 mmHg using any antihypertensive medication for 12 months, and no history of major cardiovascular events. Moonen 2015 included participants using any antihypertensive medication with a Mini‐Mental State Examination (MMSE) score of 21‐27. Walma 1997 included participants using diuretics for six months with no history of acute heart failure.

Interventions

Of six studies included, four included participants taking diuretics (Burr 1977; Maland 1983; Myers 1982; Walma 1997), and two studies included participants taking any antihypertensive medication (Langford 1984; Moonen 2015). The study by Langford 1984 included an additional intervention (diet) with participants stratified by obesity status, however, for this systematic review, we only included the groups which did not have the dietary intervention. The discontinuation plan was specified in three studies (Langford 1984; Moonen 2015; Walma 1997), as outlined in Table 1.

Outcomes

All studies reported at least one of the primary outcomes. All‐cause mortality was reported in Burr 1977; Maland 1983; Moonen 2015; and Myers 1982, with Myers 1982 also reporting cardiovascular mortality. Two studies reported myocardial infarction outcomes (Maland 1983; Moonen 2015), and three studies reported on adverse drug reactions or adverse drug withdrawal reactions (Burr 1977; Maland 1983; Myers 1982). The following secondary outcomes were reported: systolic blood pressure (SBP), five studies (Burr 1977; Maland 1983; Moonen 2015; Myers 1982; Walma 1997); stroke, three studies, (Maland 1983; Moonen 2015; Myers 1982) and success of withdrawal as measured by the ability to remain off the medication, four studies (Langford 1984; Maland 1983; Myers 1982; Walma 1997). Moonen 2015 was the only study to report on the hospitalisation and quality of life (QoL) secondary outcomes. No study reported on falls.

Funding

The trials by Burr 1977, Langford 1984 and Maland 1983 all reported pharmaceutical company support through supply of medications and identical placebos to be used in the study. It was not stated, however, what, if any, other involvement the companies had in the study concept, execution, analysis or reporting. Three studies reported nonpharmaceutical company funding (Moonen 2015; Myers 1982; Walma 1997).

Excluded studies

We excluded 33 publications (reasons for exclusion are shown in Characteristics of excluded studies). Reasons for exclusion were wrong intervention (participants were not randomised to withdraw or continue antihypertensives or withdrawal was temporary); wrong patient population (not older adults or wrong indication); wrong study design (not an RCT); one or both groups had confounding treatment (such as additional dietary or exercise intervention or other antihypertensives); clinical trial registry citation of an ineligible study due to wrong patient population; and a protocol paper of an ineligible study due to wrong intervention.

Risk of bias in included studies

See Characteristics of included studies for details of the risk of bias in the included studies; there was variable risk of bias across the studies. Figure 2 and Figure 3 show the percentages of low, unclear and high risk of bias across the different domains and the risk of bias in the different domains of the individual studies.

Figure 2
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

All studies were assessed as having low or unclear risk of selection bias. Two studies reported appropriate methods of random sequence generation and were assessed as low risk in this domain (Moonen 2015; Walma 1997). Three studies had a low risk of bias in the allocation concealment domain; Myers 1982 and Walma 1997 used hospital pharmacists to ensure allocation concealment while Moonen 2015 noted a centralised computer randomisation procedure. The majority of studies had an unclear risk of selection bias due to insufficient information.

Blinding

Four of the studies were assessed as having low risk of bias in the blinding of participants and personnel domain as they reported that the study was double‐blind and also described some method of blinding, such as using identical placebos for the withdrawal group (Burr 1977; Maland 1983; Myers 1982; Walma 1997). Burr 1977 also noted that the containers were special so as to ensure that personnel were blinded to allocation. Of these four studies, two (Maland 1983; Myers 1982) had insufficient details about blinding of assessors (unclear risk of bias) and one (Walma 1997) was assessed as having low risk of detection bias as it reported that allocation codes were not broken until assessment of the last data had been completed.

The trial by Moonen 2015 did not blind the participants and physicians to treatment group, however, the research personal assessing the outcomes were masked to allocation group. Finally, Langford 1984 didn't specify whether blinding was conducted; external physicians could restart the medication, therefore it is unlikely to have been blinded. As restarting the medication was the primary outcome, this study was assessed as having high risk of performance and detection bias.

Incomplete outcome data

Three studies were assessed as having low risk of attrition bias (Maland 1983; Moonen 2015; Walma 1997), one was unclear (Langford 1984) and the remaining two were high risk (Burr 1977; Myers 1982). Burr 1977 had unclear reporting of dropouts and inconsistent reporting of outcomes particularly for the continuation group, while in the study by Myers 1982, there were uneven dropout rates (with different reasons for dropouts between the groups) and incomplete data reported on their main outcome (BP).

Selective reporting

Only one study was assessed as having low risk of reporting bias (Walma 1997), with the others being unclear (Burr 1977; Myers 1982) or at high risk of bias (Langford 1984; Maland 1983; Moonen 2015). Two studies were assessed as at high risk of bias as outcomes appeared to be reported in unclear and/or select ways with grouping of participants not as per the originally allocated groups (Langford 1984; Maland 1983). For example, Maland 1983 reported BP results for 'reverters' (those whose BP increased to a level requiring restarting of treatment) separately to other participants in the withdrawal group. The third study assessed as being at high risk of reporting bias reported an outcome in their protocol (attached as supplementary data) which was not reported in the main manuscript (Moonen 2015).

Other potential sources of bias

No other potential sources of bias were identified in the studies.

Effects of interventions

See: Summary of findings 1 Discontinuation by no treatment/placebo of antihypertensives compared to continuation in older people

Primary outcomes

Mortality

Four studies (Burr 1977; Maland 1983; Moonen 2015; Myers 1982) reported all‐cause mortality (Analysis 1.1); Maland 1983 and Myers 1982 had up to 12 months follow‐up while Burr 1977 had 12 weeks and Moonen 2015 had 16 weeks. All four studies, with a total of 640 participants, were included in a meta‐analysis (Figure 4). The odds for all‐cause mortality were 2.08 (95% Confidence Interval (CI) = 0.79 to 5.46, I2 = 0%) of that in the discontinuation group compared to the control group. The certainty of the evidence was low, mostly due to the small number of events in the included studies and also the CI included both the null effect as well as what could be considered an appreciable harm of discontinuation (i.e. favouring continuation).

Figure 4
Forest plot of comparison: 1 Continuation vs discontinuation by no treatment/placebo of antihypertensives, outcome: 1.1 All‐cause mortality.

Forest plot of comparison: 1 Continuation vs discontinuation by no treatment/placebo of antihypertensives, outcome: 1.1 All‐cause mortality.

Myers 1982 also reported cardiovascular mortality; in the 12 month follow‐up period, two participants in the withdrawal group (n = 38) and one in the continuation group (n = 39) were reported to have died from cardiovascular causes (Analysis 1.2; very low certainty of evidence). One in each group died from heart failure and the second participant in the withdrawal group had a stroke (normal BP).

Myocardial infarction

Two studies reported the outcome of myocardial infarctions (MIs); Maland 1983 had a 12‐month follow‐up and Moonen 2015 had a 16‐week follow‐up. Both were included in the meta‐analysis which found no evidence of a difference between the groups (OR = 1.86, 95% CI = 0.19 to 17.98, I2 = 0%; Analysis 1.3). The certainty of the evidence was very low.

Adverse drug reactions and adverse drug withdrawal reactions

Meta‐analysis for this outcome was not possible due to a large variation in how this outcome was reported between studies. Maland 1983 reported that there was no difference between groups in frequency of side effects or well‐being although data for these outcomes was not shown. It was noted that in the continuation group (n = 31), one participant was removed from the study due to a complaint of sexual dysfunction and another for persistently elevated blood sugar. Reversal of a slight postural drop in SBP (observed at baseline in both groups) was reversed in the discontinuation group in the study by Myers 1982. Additionally, renal function and serum cholesterol levels improved (small but significant) in the discontinuation group (Myers 1982).

Three studies with participants taking mostly loop or thiazide‐type diuretics reported change in potassium levels. Burr 1977 reported that participants with low potassium at baseline had this reversed in the discontinuation group but not in the continuation group and Maland 1983 noted an increase in potassium in the discontinuation group. Myers 1982 reported that potassium increased in both groups (potassium supplementation was, however, prescribed at the discretion of the physician to both groups).

None of the studies reported whether or not any adverse drug withdrawal reactions occurred during the study period. However, Burr 1977 reported that the main clinical effect of withdrawal was increase in ankle oedema. In the discontinuation group (n = 41), 21 had an increase in oedema, 14 were unchanged and 6 had decreased levels with corresponding numbers in the continuation group (n = 48) of 14, 19 and 15. In most cases, the oedema increase was slight. Myers 1982, however, reported significant reductions in ankle oedema in both groups.

The certainty of the evidence for adverse drug reactions and adverse drug withdrawal reactions was determined to be very low due to very serious concern about lack of blinding, attrition bias and poor detection and reporting of outcomes and the small sample size and number of events.

Secondary outcomes

Blood pressure (BP)

Five studies (n = 767) were included in a meta‐analysis of systolic blood pressure (SBP) (Burr 1977; Maland 1983; Moonen 2015; Myers 1982; Walma 1997). A mean difference of 9.75 mmHg (95% CI = 7.33 to 12.18) in favour of continuation compared to discontinuation was found (Figure 5Analysis 1.4). The certainty of the evidence was low.

Figure 5
Forest plot of comparison: 1 Discontinuation by no treatment/placebo of antihypertensives vs Continuation, outcome: 1.4 Systolic blood pressure.

Forest plot of comparison: 1 Discontinuation by no treatment/placebo of antihypertensives vs Continuation, outcome: 1.4 Systolic blood pressure.

Substantial heterogeneity (I2 = 67%) was observed in the SBP meta‐analysis and, therefore, subgroup analyses by duration of use and class of medication were conducted according to prespecified plans. Three studies had a follow‐up of less than 12 months (up to 12 weeks (Burr 1977), 16 weeks (Moonen 2015) and six months (Walma 1997)), and two had a follow‐up of 12 months (Maland 1983; Myers 1982). This subgroup analysis was not able to explain the heterogeneity, and there were no significant differences between the mean differences in the two subgroups (Analysis 1.10; duration of follow‐up < 12 months: mean difference = 9.17 mmHg, 95% CI = 6.62 to 11.71; duration of follow‐up ≥ 12 months: mean difference = 15.30 mmHg, 95% CI = 7.48 to 23.12, P = 0.14). Subgroup analysis by drug class (diuretics (Burr 1977; Maland 1983; Myers 1982; Walma 1997) versus other (Moonen 2015)) was also not able to explain the heterogeneity and there was no difference between subgroups (Analysis 1.11).

The same five studies (n = 768) were also included in a meta‐analysis of change in diastolic blood pressure (DBP) (Burr 1977; Maland 1983; Moonen 2015; Myers 1982; Walma 1997). A mean difference of 3.50 mmHg (95% CI = 1.82 to 5.18) was found in favour of continuation (Analysis 1.5) with a low certainty of evidence. Moderate heterogeneity was found (I2 = 47%) and so subgroup analysis by duration of follow‐up and drug class was conducted. The heterogeneity in the DBP analysis might be explained by duration of follow‐up. The subgroup of < 12 months duration of follow‐up had no heterogeneity, although heterogeneity remained in the studies with ≥ 12 months duration (Figure 6, Analysis 1.12). There was a significant difference between the studies with a follow‐up of < 12 months compared to ≥ 12 months (mean difference = 2.68 mmHg, 95% CI = 0.87 to 4.49 and mean difference = 8.70 mmHg, 95% CI = 4.15 to 13.25 respectively, P = 0.02). Subgroup analysis by type of antihypertensive drug didn't explain the heterogeneity and there was no significant difference between these subgroups (Analysis 1.13).

Figure 6
Forest plot of comparison: 1 Discontinuation by no treatment/placebo of antihypertensives vs Continuation, outcome: 1.12 Diastolic blood pressure sub grouped on duration of study.

Forest plot of comparison: 1 Discontinuation by no treatment/placebo of antihypertensives vs Continuation, outcome: 1.12 Diastolic blood pressure sub grouped on duration of study.

Hospitalisation

Hospitalisation was reported as an outcome in a single study involving participants with mild cognitive defects (Moonen 2015). There were nine participants in the withdrawal group (n = 199) and 10 participants in the continuation group (n = 186) hospitalised during the 6‐week follow‐up (Analysis 1.6; OR = 0.83, 95% CI = 0.33 to 2.10). These results were noted to not include the participants who had died or experienced a vascular event. There was a low certainty of evidence due to serious concerns about the sample size and number of events.

Stroke (fatal and non‐fatal, ischaemic and haemorrhagic, transient ischaemic attack)

Three studies with durations of 16 weeks to 12 months had stroke as an outcome (Maland 1983; Moonen 2015; Myers 1982) with a total of 524 participants included in a meta‐analysis (Analysis 1.7). There was no evidence of a difference between discontinuation and continuation groups (OR = 1.44, 95% CI = 0.25 to 8.35). The certainty of the evidence was determined to be low due to serious concerns about indirectness and imprecision.

Success (rate) of withdrawal from antihypertensive drugs

Four studies reported the success of withdrawal as measured by the ability to remain off the medication. Three studies (Maland 1983; Myers 1982; Walma 1997; n = 341) were included in the meta‐analysis which found an OR of 3.23 (95% CI 1.86 to 5.61, I2 = 71.42%) in favour of continuation (Analysis 1.8). There was low certainty due to heterogeneity and indirectness.

Myers 1982 reported the number of participants that had to be withdrawn from the study due to clinically important hypertension (two consecutive readings above 180/110 mmHg) or heart failure in the 12‐month follow‐up period. Four out of 38 (10.5%) participants in the discontinuation group were withdrawn (two for hypertension and two for heart failure) while six out of 39 (15.4%) participants in the continuation group were withdrawn (all due to heart failure). Maland 1983 also had a 12‐month follow‐up period and reported the number of participants who had to be restarted on their antihypertensive medication due to hypertension. The assessment was made using predefined criteria: average DBP ≥ 105 mmHg at any one visit, DBP 96 to 104 mmHg at any two visits, DBP > 90 mmHg at any five visits; or DBP > 90 mmHg at every visit in the first 24 weeks. Eight out of 31 (25.8%) and one out of 31 (3.2%) discontinuation and continuation participants were removed from the study due to hypertension, respectively. Participants in the discontinuation group who had to restart their medication due to hypertension were slightly older and had slightly higher blood pressure before treatment than those who didn't, however, the differences were not statistically significant. Walma 1997 was the largest study with this outcome, but had a shorter follow‐up of 6 months and successful withdrawal from therapy was the primary outcome. Predefined criteria for re‐initiation of diuretic therapy were heart failure (score of four or greater) or hypertension (an average of three measurements on separate occasions of SBP > 180 mmHg or DBP > 100 mmHg). In the discontinuation group, 34 (25 for heart failure, nine for hypertension) out of 102 (33.3%) required re‐initiation of therapy and, in the continuation group, nine (four for heart failure and five for hypertension) out of 100 (9%) met the criteria. This study also reported that 16 and four participants in the discontinuation and continuation groups had their diuretic restarted by the doctor for other reasons (such as increased shortness of breath).

The study by Langford 1984 could not be included in the meta‐analysis as no data on the outcome could be found for the continuation group. This study reported the percentage of participants in the discontinuation group who restarted the antihypertensive medication or had a terminating event (physician outside of the study restarted the medication, stroke, new MI, heart failure or elevated creatinine level). The study criteria for restarting the medication was DBP 95 to 99 mmHg on three occasions within three months, DBP 100 to 104 mmHg on two occasions within one month or DBP ≥ 105 mmHg at any time. Langford 1984 reported that 35.3% and 45.0% in the obese and nonobese discontinuation groups (without dietary intervention) respectively remained withdrawn from their medication(s) after 56 weeks.

Quality of life (QoL)

One study (n = 356, participants with mild cognitive defects) reported change in quality of life (QoL) in the discontinuation and continuation groups using the Cantril Ladder (Moonen 2015) (single item scale, range from 1‐10 with higher scores indicating better QoL). There was no significant difference in change in score after 16 weeks between the discontinuation and continuation groups (−0.09, 95% CI = −0.34 to 0.16; P = 0.46). It was determined that there was low certainty of evidence for this outcome due to risk of bias and small sample size (single study).

Falls

None of the included studies reported outcomes or data on the incidence, rate or risk of falls.

Discussion

Summary of main results

Our systematic review identified six RCTs on antihypertensive discontinuation in older people. Based on currently available evidence, discontinuation of antihypertensives has no effect on all‐cause mortality, MI, or stroke, compared with continuation; however, there is low or very low certainty in these results. Eligible studies were generally small and had short‐term follow‐up with few numbers of events. The meta‐analysis results for these outcomes also had wide confidence intervals, which included both the null hypothesis (no difference) as well as an appreciable benefit in favour of continuation. Therefore, additional studies may change these results. We found no evidence that antihypertensive discontinuation increased the risk of adverse drug events, with some indication of a reversal of adverse drug reactions with discontinuation; however, eligible studies did not assess adverse drug withdrawal reactions specifically and, in general, all reporting of adverse drug events was very poor. It should be noted that the review and synthesis of adverse drug reactions and adverse drug withdrawal reactions did not include outcomes that were considered separately, for example, restarting of therapy due to increased BP or other clinical reasons. We did find evidence surrounding the effect of discontinuation on blood pressure and restarting of antihypertensives, however, again the certainty of evidence is low. Discontinuation led to an increase in both SBP and DBP compared to continuation, though there was heterogeneity in these estimates which could not be fully explained. Further, the proportion of persons withdrawing from studies due to worsening hypertension or heart failure (based on predefined criteria) leading to subsequent restarting of antihypertensive medications was higher in the discontinuation arm. Finally, we found that there was no effect of antihypertensive discontinuation on quality of life or hospitalisations (low certainty of evidence); however, these outcomes were only assessed in one study each.

Overall, the limited available evidence suggests that antihypertensive discontinuation in older people may have no effect on clinically‐important outcomes such as mortality or cardiovascular events, with low certainty in this result. However, discontinuation may lead to an increase in blood pressure, requiring some patients to restart therapy. Between 10.5% and 33.3% of participants in the discontinuation group required restarting of therapy due to blood pressure or other clinical criteria compared to rates of 9% to 15% in the continuation group. A slightly higher rate of restarting (35.3% to 45.0%) was reported in one of the studies which didn't report the rate in the continuation group.

Overall completeness and applicability of evidence

There are a few factors which increase the applicability of the evidence to the population of interest, however, overall there is a lack of completeness and concern about the applicability of the evidence to the majority of older adults using antihypertensives for hypertension or primary prevention of CVD.

All included studies had inclusion criteria that mimic situations where discontinuation of the antihypertensive could be considered, that is, older adults with controlled blood pressure. However, there are other scenarios where discontinuation of antihypertensives would be considered that are not reflected in these studies, for instance, in situations where the individual has low BP, a postural drop in BP, or where they are potentially suffering from harm such as falls. Additionally, there was significant variability in the inclusion and exclusion criteria as to what they defined controlled blood pressure to be. For example, Walma 1997 excluded participants if their SBP was greater than 180 mmHg, while Moonen 2015 used a more conservative exclusion criteria of greater than 140 mmHg.

The conditions of RCTs limits the applicability of our findings. Specifically, in clinical practice, discontinuation of antihypertensive medications would be a result of shared decision‐making, taking into account the values and preferences of the patient. Additionally, after discontinuation, blood pressure can be monitored and the medication would be restarted if the blood pressure rose above the acceptable level (for the individual). Additionally, only three (Langford 1984; Moonen 2015; Walma 1997) of the six studies reported tapering the dose of the medication prior to discontinuation and one only required tapering of beta‐blockers and not other antihypertensives (Moonen 2015). Tapering would likely be conducted in clinical practice to determine the lowest effective dose of the medication (Reeve 2014b; Scott 2019). While the studies had criteria for definition of relapse of hypertension and study withdrawal, this was not consistent among the studies.

Another limitation to the applicability of the evidence is that five out of the six studies were conducted over 20 years ago. Standards of treatment, population risk factors (e.g. smoking) and recommendations about non‐drug approaches have changed and there has been a reduction in cardiovascular mortality over this time period (Mensah 2017). There has also been an increase in the number of the oldest old, that is, those who are over 85 years old. These changes make it difficult to determine the applicability to the current population of older adults as the net treatment effect may have been altered over the intervening two decades. Similarly, over the past 20 years, there has been an increase in the prevalence of polypharmacy and this is now commonplace in older adults (Oktora 2019; Page 2019). None of the included studies reported other medication use or noncardiovascular morbidity profiles of participants at baseline and so it is not possible to determine the applicability of the results to older adults with polypharmacy and multimorbidity. Additionally, no study measured or reported the frailty of participants. As frailty has recently been recognised as a key issue in relation to both the likely benefits and harms of antihypertensive use in older adults, ensuring assessment of frailty using validated tools is essential for future research in this area (Scott 2019).

Limited evidence was found for the majority of clinical outcomes. Cardiovascular mortality, hospitalisation, and quality of life were only reported in a single study each and none of the studies reported falls as an outcome. Several of the studies reported all‐cause mortality and stroke, however, as the confidence intervals crossed both the null and a clinically important magnitude, we cannot make any firm conclusions about these outcomes. Additionally, there was little detail and apparently insufficient methods to capture adverse drug reactions and adverse drug withdrawal reactions.

Finally, four of the studies examined discontinuation of diuretics only and the remaining two included discontinuation of any antihypertensive medication. The results should be interpreted with caution for nondiuretic antihypertensive medications. We also did not identify any studies on the feasibility or outcomes of dose reduction of antihypertensives.

Quality of the evidence

The certainty of the evidence was judged to be low or very low for all the outcomes considered in this review. Therefore, there is uncertainty in the evidence of the effect of antihypertensive withdrawal on outcomes overall. The reasons for downgrading were risk of bias, inconsistency, indirectness, and imprecision.

While most studies were published prior to publication of the CONSORT statement (Moher 2009), they were identified to have methodological and reporting limitations. Only one RCT was assessed as having low risk of bias for all domains (Walma 1997). In four out of the six studies, there was insufficient information on the random sequence generation. The risk of bias from allocation concealment was judged as unclear in half of the studies. Two studies were judged as having a high level of potential performance bias due to lack of blinding of participants and personnel. Reporting on blinding of outcome assessment was judged as unclear or leading to high risk of bias in most studies. Two studies had high risk of bias for incomplete outcome data (high numbers and unbalanced dropouts). Selective reporting was judged to be unclear or leading to high risk of bias in five of the six included studies.

Potential biases in the review process

There are a number of limitations to our review process. Firstly, the importance of deprescribing and need to explore the benefits and harms is a relatively new concept. As such, data on the feasibility and outcomes in early studies were poorly reported. Additionally, while we used a variety of different search terms for deprescribing/withdrawal, it is possible that early studies reporting such results may not have used these terms and the search strategy we used has not been validated. Therefore, our search strategy may not have identified all relevant studies.

Few studies overall were included in this review; this is likely in part to result from our strict inclusion/exclusion criteria. It is important to consider that we excluded studies which included any additional intervention in either group, for example, dietary or exercise intervention in those that discontinued which would likely better reflect ideal practice. However, in older adults, the ability to conduct and adhere to such lifestyle interventions might be limited in real world practice. We also only included studies that withdrew all antihypertensive medications; the results of this review are not relevant to those taking multiple antihypertensives, where one is being considered for discontinuation.

We aimed to only include studies where the medication was used for hypertension or primary prevention of cardiovascular disease. However, many of the studies reported criteria for restarting that were related to heart failure. Therefore, it is possible that the populations in the studies did not truly satisfy our intended population.

Agreements and disagreements with other studies or reviews

This is the first systematic review and meta‐analysis to investigate whether withdrawal of antihypertensive medications in older people with hypertension is feasible, and evaluate the effects of withdrawal on multiple outcome measures including mortality, cardiovascular outcomes, hypertension and quality of life. There have been a limited number of previous reviews which have aimed to assess the efficacy and safety of withdrawing antihypertensive medications in older people. Froom 1997 conducted a nonsystematic review of the effect of withdrawal of antihypertensive medications. In subgroup analysis of the six observational studies limited to older people, there was an average success rate of 26.2% for periods of two or more years, however, the authors did not report on specific outcome measures other than success of discontinuation. A systematic review by Iyer 2008 aimed to search for clinical studies of the benefits and harms of withdrawing a range of specific classes of medications in older people, not just antihypertensives. They included and reported on nine open‐label, prospective observational studies of antihypertensive withdrawal in older people. No RCTs were included. Iyer 2008 reported that between 20% and 85% of participants did not recommence antihypertensive medications over a period of 4 to 260 weeks. No significant withdrawal syndromes were noted and the major reason for recommencing antihypertensive therapy was a gradual increase in blood pressure and, less commonly, heart failure. Similar reasons for restarting antihypertensive medications were identified in our review (blood pressure and heart failure), however, the proportion restarting in the discontinuation arms was generally lower in the RCTs included in our review (10.5% to 45.0%) than reported by Iyer 2008.

In another systematic review not limited to antihypertensives, Page 2016 aimed to determine whether deprescribing is feasible, and the impact on mortality and health outcomes in older adults. The review included a wider range of study designs; 132 randomised and nonrandomised studies across all settings and medication classes. The authors included 13 studies in the antihypertensive class (four RCTs, one case control, one historical cohort and seven before and after studies). Similar to the results of our review, they reported increases in systolic and diastolic BP, but no statistically significant difference in mortality.

Van der Wardt 2017 conducted a systematic review of withdrawal of antihypertensives to examine success and safety, however, this review was not limited to older adults or to RCTs. They included 66 articles and concluded that approximately one‐quarter of people taking an antihypertensive could withdraw the medication without return of hypertension. Another systematic review not restricted to older adults concluded that patients with low BP, taking lower doses of antihypertensives, fewer number of antihypertensives and those motivated to implement lifestyle changes were more likely to remain normotensive after antihypertensive medication withdrawal (Nelson 2001).

A previous Cochrane review focused on the effect of withdrawal of antihypertensives on cognition (Jongstra 2016). They identified two studies (one of which was also included in our review (Moonen 2015)) and reported that, while there was a signal of a positive effect on cognition with withdrawal, there was too much uncertainty to make any strong conclusions. Previous nonRCT interventional studies have aimed to reduce the risk of falls through withdrawal of 'fall risk increasing drugs', which included antihypertensives. There have been conflicting results as to whether there is a benefit or not, however, this may have depended on how many and which medications were successfully withdrawn (Boye 2017; Van Der Velde 2007).

All these reviews conclude that deprescribing of antihypertensive medications (in those taking the medication for hypertensive or primary prevention of CVD) is likely feasible in a proportion of older adults, although the proportion in which it is likely to be successful varies. However, there is uncertainty in the likely benefit and harms of antihypertensive withdrawal.

Study flow diagram.

Figuras y tablas -
Figure 1

Study flow diagram.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figuras y tablas -
Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figuras y tablas -
Figure 3

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Forest plot of comparison: 1 Continuation vs discontinuation by no treatment/placebo of antihypertensives, outcome: 1.1 All‐cause mortality.

Figuras y tablas -
Figure 4

Forest plot of comparison: 1 Continuation vs discontinuation by no treatment/placebo of antihypertensives, outcome: 1.1 All‐cause mortality.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Forest plot of comparison: 1 Discontinuation by no treatment/placebo of antihypertensives vs Continuation, outcome: 1.4 Systolic blood pressure.

Figuras y tablas -
Figure 5

Forest plot of comparison: 1 Discontinuation by no treatment/placebo of antihypertensives vs Continuation, outcome: 1.4 Systolic blood pressure.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Forest plot of comparison: 1 Discontinuation by no treatment/placebo of antihypertensives vs Continuation, outcome: 1.12 Diastolic blood pressure sub grouped on duration of study.

Figuras y tablas -
Figure 6

Forest plot of comparison: 1 Discontinuation by no treatment/placebo of antihypertensives vs Continuation, outcome: 1.12 Diastolic blood pressure sub grouped on duration of study.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 1: All‐cause mortality

Figuras y tablas -
Analysis 1.1

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 1: All‐cause mortality

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 2: Cardiovascular mortality

Figuras y tablas -
Analysis 1.2

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 2: Cardiovascular mortality

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 3: Myocardial infarction (fatal and nonfatal)

Figuras y tablas -
Analysis 1.3

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 3: Myocardial infarction (fatal and nonfatal)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 4: Systolic blood pressure

Figuras y tablas -
Analysis 1.4

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 4: Systolic blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 5: Diastolic blood pressure

Figuras y tablas -
Analysis 1.5

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 5: Diastolic blood pressure

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 6: Hospitalisation

Figuras y tablas -
Analysis 1.6

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 6: Hospitalisation

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 7: Stroke (fatal + nonfatal + TIA)

Figuras y tablas -
Analysis 1.7

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 7: Stroke (fatal + nonfatal + TIA)

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 8: Sucess rate ‐ withdrawal/resumption due to hypertension or other clinical reason

Figuras y tablas -
Analysis 1.8

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 8: Sucess rate ‐ withdrawal/resumption due to hypertension or other clinical reason

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 9: Quality of life

Figuras y tablas -
Analysis 1.9

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 9: Quality of life

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 10: Systolic blood pressure subgrouped on duration

Figuras y tablas -
Analysis 1.10

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 10: Systolic blood pressure subgrouped on duration

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 11: Systolic blood pressure subgrouped on drug type

Figuras y tablas -
Analysis 1.11

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 11: Systolic blood pressure subgrouped on drug type

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 12: Diastolic blood pressure subgrouped on duration

Figuras y tablas -
Analysis 1.12

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 12: Diastolic blood pressure subgrouped on duration

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 13: Diastolic blood pressure subgrouped on drug type

Figuras y tablas -
Analysis 1.13

Comparison 1: Discontinuation by no treatment/placebo of antihypertensives vs Continuation, Outcome 13: Diastolic blood pressure subgrouped on drug type

Ir a la figura de la revisiónAbrir en una pestaña nueva
Summary of findings 1. Discontinuation by no treatment/placebo of antihypertensives compared to continuation in older people

Discontinuation by no treatment/placebo of antihypertensives compared to continuation in older people

Patient or population: Older adults, 50 years and older
Setting: All settings
Intervention: Discontinuation by no treatment/placebo of antihypertensives
Comparison: Continuation

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

No. of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with Continuation

Risk with Discontinuation by no treatment/placebo of antihypertensives

All‐cause mortality
follow‐up: range 12 weeks to 12 months

Study population

OR 2.08
(0.79 to 5.46)

630
(4 RCTs)

⊕⊕⊝⊝
LOW 1 2 3 4

19 per 1,000

40 per 1,000
(15 to 98)

Moderate

26 per 1,000

52 per 1,000
(20 to 126)

Myocardial infarction (fatal and nonfatal)
follow‐up: range 16 weeks to 12 months

Study population

OR 1.86
(0.19 to 17.98)

447
(2 RCTs)

⊕⊝⊝⊝
VERY LOW 5 6

5 per 1,000

9 per 1,000
(1 to 77)

Moderate

3 per 1,000

5 per 1,000
(1 to 46)

Hospitalisation
follow‐up: 16 weeks

Study population

OR 0.83
(0.33 to 2.10)

385
(1 RCT)

⊕⊕⊝⊝
LOW 7

54 per 1,000

45 per 1,000
(18 to 107)

Stroke (fatal + nonfatal + TIA)
follow‐up: range 16 weeks to 12 months

Study population

OR 1.44
(0.25 to 8.35)

524
(3 RCTs)

⊕⊕⊝⊝
LOW 38

8 per 1,000

11 per 1,000
(2 to 62)

Moderate

5 per 1,000

8 per 1,000
(1 to 43)

Systolic blood pressure
follow‐up: range 12 weeks to 12 months

The mean systolic blood pressure ranged from 123 to 145 mmHg 9

MD 9.75 mmHg higher
(7.33 higher to 12.18 higher)

‐

767
(5 RCTs)

⊕⊕⊝⊝
LOW 10 11 12

Diastolic blood pressure: range 12 weeks to 12 months

The mean diastolic blood pressure ranged from 70‐95 mmHg 9

MD 3.5 mmHg higher
(1.82 higher to 5.18 higher)

‐

768
(5 RCTs)

⊕⊕⊝⊝
LOW 10 12 13

Adverse drug reactions and adverse drug withdrawal reactions (adverse reactions)
follow‐up: range 12 weeks to 12 months

Overall there was some reversal of adverse drug reactions in the discontinuation group, otherwise no change reported.

One study reported no difference in frequency of side effects although data were not shown.

One study reported reversal of slight postural drop and improvement in renal function and serum cholesterol in the discontinuation but not the continuation group.

One study reported that more participants experienced ankle oedema in the discontinuation than the continuation group, however, statistical significance was not reported.

All three studies reported changes in potassium in the discontinuation group.

‐

245
(3 RCTs)

⊕⊝⊝⊝
VERY LOW 14 15

Falls ‐ not reported

None of the included studies reported on falls or falls risk.

‐

‐

‐

*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;

GRADE Working Group grades of evidence
High certainty: We are very confident that the true effect lies close to that of the estimate of the effect
Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different
Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect
Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

1 Not downgraded for risk of bias: some concern about attrition bias in 2 of the studies (high and uneven drop out rates), not enough to assess as serious but considered when downgrading in other areas.

2 Not downgraded for inconsistency but considered when downgrading other areas: some concern about heterogeneity due to the difference in mortality rates between Myers 1982 and Maland 1983 but likely explained by differences in populations (Maland's participants were older and recruited from geriatric institutions).

3 Downgraded one level for imprecision, the total number of events was very low ; additionally, the CI included the null effect and appreciable benefit favouring continuation.

4 Downgraded one level for indirectness due to concern about the age of the majority of studies relevant for this outcome ‐ standards of care and recommendations for treatment have significantly changed over the past 35 years. Downgrading in this category also took into account the large concern about imprecision (which was downgraded 1, but had considered downgrading 2 steps) as well as potential risk of bias and inconsistency.

5 Downgraded one level for risk of bias as neither study had blinding of participants or physicians; possible that diagnosis of MI could have been influenced by knowledge of intervention.

6 Downgraded two levels for imprecision as the total number of events was very low; additionally, the CI included the null effect and appreciable benefit favouring continuation.

7 Downgraded two levels for imprecision as the total number of events was very low; additionally, the CI included the null effect and appreciable benefit favouring discontinuation.

8 Downgraded one level for indrectness due to concern about the age of the majority of studies relevant for this outcome ‐ standards of care and recommendations for treatment have significantly changed over the past 35 years. Downgrading in this category also took into account the large concern about imprecision (which was downgraded 1, but had considered downgrading 2 steps)..

9 Range of BP at end of follow‐up period in continuation group of included studies.

10 Downgraded one level for risk of bias due to concern about lack of blinding of participants and physicians in two of the studies, uneven dropouts in two of the studies and reporting bias in one of the studies.

11 Downgraded one level for inconsistency due to substantial heterogeneity, subgroup analyses based on duration of follow‐up and class of medication was not able to explain heterogeneity, and no other cause identified.

12 Some concern about indirectness as BP is a surrogate marker, also the majority of studies are more than 35 years old. Did not downgrade in this category, however contributed to the decision to downgrade in the 'Risk of bias' category.

13 Downgraded one level for inconsistency; duration of follow‐up was able to explain some of the inconsistency, however, heterogeneity remained in the subgroup with a duration of follow‐up 12 months or greater.

14 Downgraded two levels for risk of bias due to very serious concern about lack of blinding, attrition bias and poor detection and reporting of outcome.

15 Downgraded one level for imprecision due to small sample size and number of events.

TIA: Transient ischaemic attack
Figuras y tablas -
Summary of findings 1. Discontinuation by no treatment/placebo of antihypertensives compared to continuation in older people
Ir a la tabla de la revisión
Table 1. Summary of studies

Study, country

Setting

Study duration

Antihypertensive medication class withdrawn

Discontinuation plan

Burr 1977, Wales

Long‐stay geriatric wards

12 months

Diuretic

Not specified

Langford 1984, USA

Primary care

56 weeks

Any antihypertensive medication

Participants were withdrawn from therapy using a standardised step‐down withdrawal programme taking from 2 to 8 weeks, depending on the number and dosage of drugs at entry. The diuretic was the last agent withdrawn.

Drug withdrawal took place in a stepped fashion with the highest step drug being removed first. Target time for withdrawal was 6 weeks.

Maland 1983, USA

Community

12 months

Diuretic

Not specified

Moonen 2015, Netherlands

128 general practices

16 weeks

β‐Blocker, diuretic, angiotensin‐converting enzyme inhibitor, angiotensin receptor blocker or calcium channel blocker.

The discontinuation of antihypertensive treatment was performed by the participant’s physician according to an algorithm composed by the investigators.

All physicians were instructed to withdraw antihypertensive treatment until a maximum increase of 20 mmHg in SBP was reached.

The physician monitored BP every week until no further changes in antihypertensive treatment were made

Myers 1982, Canada

Geriatric institution

12 months

Diuretic

Not specified

Walma 1997, Netherlands

8 general practices

6 months

Diuretic

Participants with frusemide dosages of 40 or 80 mg/day went through a dose‐halving regimen of one and two weeks, respectively, to prevent severe rebound effects.

Dose halving was started immediately after randomisation and was performed double‐blind.

BP: Blood pressure
SBP: Systolic blood pressure

Figuras y tablas -
Table 1. Summary of studies
Ir a la tabla de la revisión
Comparison 1. Discontinuation by no treatment/placebo of antihypertensives vs Continuation

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1.1 All‐cause mortality Show forest plot

4

630

Peto Odds Ratio (Peto, Fixed, 95% CI)

2.08 [0.79, 5.46]

1.2 Cardiovascular mortality Show forest plot

1

77

Peto Odds Ratio (Peto, Fixed, 95% CI)

2.04 [0.21, 20.19]

1.3 Myocardial infarction (fatal and nonfatal) Show forest plot

2

447

Peto Odds Ratio (Peto, Fixed, 95% CI)

1.86 [0.19, 17.98]

1.4 Systolic blood pressure Show forest plot

5

767

Mean Difference (IV, Fixed, 95% CI)

9.75 [7.33, 12.18]

1.5 Diastolic blood pressure Show forest plot

5

768

Mean Difference (IV, Fixed, 95% CI)

3.50 [1.82, 5.18]

1.6 Hospitalisation Show forest plot

1

385

Peto Odds Ratio (Peto, Fixed, 95% CI)

0.83 [0.33, 2.10]

1.7 Stroke (fatal + nonfatal + TIA) Show forest plot

3

524

Peto Odds Ratio (Peto, Fixed, 95% CI)

1.44 [0.25, 8.35]

1.8 Sucess rate ‐ withdrawal/resumption due to hypertension or other clinical reason Show forest plot

3

341

Peto Odds Ratio (Peto, Fixed, 95% CI)

3.23 [1.86, 5.61]

1.9 Quality of life Show forest plot

1

356

Mean Difference (IV, Fixed, 95% CI)

‐0.10 [‐0.35, 0.15]

1.10 Systolic blood pressure subgrouped on duration Show forest plot

5

767

Mean Difference (IV, Fixed, 95% CI)

9.75 [7.33, 12.18]

1.10.1 Less than 12 months

3

647

Mean Difference (IV, Fixed, 95% CI)

9.17 [6.62, 11.71]

1.10.2 12 months or longer

2

120

Mean Difference (IV, Fixed, 95% CI)

15.30 [7.48, 23.12]

1.11 Systolic blood pressure subgrouped on drug type Show forest plot

5

767

Mean Difference (IV, Fixed, 95% CI)

9.75 [7.33, 12.18]

1.11.1 Diuretics

4

411

Mean Difference (IV, Fixed, 95% CI)

10.85 [7.92, 13.78]

1.11.2 Other

1

356

Mean Difference (IV, Fixed, 95% CI)

7.40 [3.10, 11.70]

1.12 Diastolic blood pressure subgrouped on duration Show forest plot

5

768

Mean Difference (IV, Fixed, 95% CI)

3.50 [1.82, 5.18]

1.12.1 Less than 12 months

3

646

Mean Difference (IV, Fixed, 95% CI)

2.68 [0.87, 4.49]

1.12.2 12 months or longer

2

122

Mean Difference (IV, Fixed, 95% CI)

8.70 [4.15, 13.25]

1.13 Diastolic blood pressure subgrouped on drug type Show forest plot

5

768

Mean Difference (IV, Fixed, 95% CI)

3.50 [1.82, 5.18]

1.13.1 Diuretics

4

412

Mean Difference (IV, Fixed, 95% CI)

4.42 [2.03, 6.81]

1.13.2 Other

1

356

Mean Difference (IV, Fixed, 95% CI)

2.60 [0.24, 4.96]
Figuras y tablas -
Comparison 1. Discontinuation by no treatment/placebo of antihypertensives vs Continuation
Ir a la tabla de la revisión