Fitness training for cardiorespiratory conditioning after traumatic brain injury

Summary of findings for the main comparison. Cardiorespiratory exercise programmes prescribed alone compared with a non‐exercise intervention or no intervention for people after traumatic brain injury

Cardiorespiratory exercise programmes prescribed alone compared with usual care, a non‐exercise intervention, or no intervention for people after traumatic brain injury
Patient or population: people with traumatic brain injury of any age or severity Setting: any setting, including inpatient, outpatient, community and home Intervention: exercise using large muscle groups prescribed at least three times a week for a duration of at least 20 minutes at an intensity that represented a clear attempt to improve cardiorespiratory fitness Comparison: usual care, a non‐exercise intervention, or no intervention
Outcomes	*Illustrative comparative risks (95% CI)**		No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	No fitness training	Fitness training
Cardiorespiratory fitness ‐ peak power output (W) ‐ end of intervention	The mean peak power output ranged across control groups from 85 W to 145 W	The mean peak power output in the intervention groups was on average 35 W higher (95% CI 3 to 68)	67 (3 studies)	⊕⊕ low^a,b	The MCID has not been determined for this outcome in this or a similar population, however in people with severe COPD, the MCID is 4 W (Puhan 2011).
Depression ‐ SMD ‐ end of intervention	The mean depression score ranged across control groups from 1.1 to 21.2 points on various depression scales	The SMD between groups was on average 0.43 lower in the intervention group (95% CI 0.92 to ‐0.06)	220 (4 studies)	⊕ very low ^a,b,c	An SMD of 0.43 represents a difference between groups in scores on the depression subscale of the Hospital Anxiety Depression Scale (0 to 21) of a 1.7 point reduction in depression with the 95% confidence intervals indicating an effect between a 3.7 reduction in depression to a 0.2 point increase in depression. Post hoc analysis pooling the data from the 3 studies that included subacute and chronic participants at home, the SMD indicated a moderate reduction in depression with fitness training with the 95% CIs indicating between a large and small reduction in depression with fitness training compared to the control interventions (SMD ‐0.61, 95% CI ‐1.1 to ‐0.11; 3 studies, 178 participants; low‐quality evidence).
Cognition	See comment	See comment	111 (2 studies)	See comment	Data could not be pooled because the outcome measures from both studies did not measure the same domains of cognition. There were no immediate or long‐term effects in either study.
Fatigue ‐ SMD ‐ end of intervention	The mean fatigue score ranged among control groups from 1.3 to 6.6 points on various fatigue scales	The SMD between groups was on average 0.32 lower in the intervention group (95% CI 0.90 to ‐0.26)	130 (3 studies)	⊕ very low ^a,b,c	An SMD of 0.32 represents a difference between groups in scores on a VAS fatigue scale (0 to 10) of approximately 0.9 points (95% CI ‐0.7 to 2.6), with a lower score indicating a reduction in fatigue.
Community integration	See comment	See comment	76 (1 study)	See comment	Data only available from one study; no clear evidence of effect.
Safety	See comment	See comment	183 (5 studies)	See comment	Trialists reported no adverse events occurred during the intervention phase in these studies.
The corresponding risk* (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; COPD: chronic obstructive pulmonary disease; MCID: minimal clinically important difference; SMD: standardised mean difference; VAS: visual analogue scale; W: watts
GRADE Working Group grades of evidence High quality: further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: we are very uncertain about the estimate.
^a Downgraded once because of risk of bias. ^b Downgraded once because of imprecision of results. ^c Downgraded once because of heterogeneity among included studies.

Antecedentes

available in

Descripción de la afección

El traumatismo craneoencefálico (TCE) es una causa principal de muerte y limitación a largo plazo de la actividad a nivel global (Hyder 2007). La incidencia informada del TCE varía de 108 a 392 por 100 000 habitantes (Abelson‐Mitchell 2008), y la mayoría de estas lesiones se clasifica como leve (Hyder 2007). Aunque el TCE tiene una incidencia inferior a algunos otros trastornos, como el accidente cerebrovascular, principalmente afecta a los pacientes en el mejor momento de sus vidas. La prevalencia y el costo económico y social del TCE son extensos, con un costo total directo e indirecto en los EE.UU. de entre USD 34 000 000 000 y 76 000 000 000 anualmente (Ma 2014). Además, la Organización Mundial de la Salud prevé que la incidencia del TCE crecerá debido a los accidentes automovilísticos en los países en desarrollo y a los conflictos militares en curso (Ma 2014). Por lo tanto, es importante optimizar la salud de los pacientes con TCE durante toda su vida.

Descripción de la intervención

El entrenamiento del estado cardiorrespiratorio es un método de optimización de la salud. El entrenamiento físico es un tipo de actividad física durante el tiempo libre que es planificada, estructurada, repetitiva e intencional en el sentido que tiene como objetivo la mejoría o el mantenimiento de uno o más componentes del estado físico (Caspersen 1985). El American College of Sports Medicine ha publicado guías de ejercicios para los pacientes con una lesión cerebral; las mismas recomiendan hacer ejercicio a una frecuencia de tres a cinco veces por semana, a una intensidad de un 40% a un 70% de la captación máxima de oxígeno, o una 13/20 Rating of Perceived Exertion (RPE), y durante un tiempo de 20 a 60 minutos con una modalidad apropiada de ejercicio (p.ej. caminata, natación, ciclismo) que dependerá de la capacidad física del individuo (Palmer‐McLean 2009).

De qué manera podría funcionar la intervención

La actividad física regular y un alto nivel de estado físico se asocian con un riesgo reducido de morbilidad y mortalidad a nivel global (Lee 2012; Williams 2001). Sin embargo, se ha demostrado que los pacientes con TCE son menos activos físicamente que la población sana (Hassett 2015) y presentan un bajo nivel de estado físico (Hassett 2016). En una revisión narrativa reciente realizada por el equipo que propuso un modelo causal de la reducción del estado físico en los pacientes con TCE, se agruparon los datos de 11 estudios que medían la captación máxima de oxígeno en 234 adultos con TCE (Hassett 2016). La captación máxima de oxígeno promedio fue de 27,2 ml·kg^‐1·min^‐1, que está bastante por debajo de los niveles de estado físico más bajos para los adultos de edad y sexo similar (ACSM 2014). Las razones por las que los pacientes con TCE pueden realizar menos actividades físicas son multifactoriales y pueden incluir deficiencias físicas y cognitivas y una respuesta psicológica a su lesión, así como barreras ambientales y sociales (Hassett 2016). Por ejemplo, puede ser difícil salir a caminar debido a una deficiencia física con trastornos del equilibrio o a una barrera ambiental, como un sendero con irregularidades. La deficiencia cognitiva con disminución del movimiento y el trastorno emocional con depresión también pueden disuadir a las personas con TCE de iniciar cualquier actividad y querer salir de su hogar para ir a pasear. Por tanto, es necesario investigar las intervenciones con entrenamiento físico que aumentan la actividad física de los pacientes que han presentado un TCE.

Los programas de entrenamiento físico en individuos con un estado físico deficiente ‐ pero saludables ‐ han demostrado mejorías en el estado cardiorrespiratorio de cerca de un 12% para los programas de intensidad leve (24% a 39% de la frecuencia cardíaca de reserva) a moderada (40% a 59% de la frecuencia cardíaca de reserva) (Swain 2005). En los pacientes sanos, el entrenamiento físico mejora el estado cardiorrespiratorio a través de la mejoría de los determinantes tanto centrales como periféricos (Rowell 1974). A nivel central, el entrenamiento del estado cardiorrespiratorio de suficiente intensidad aumenta el volumen sistólico y disminuye la frecuencia cardíaca en respuesta a una carga submáxima fija; y periféricamente, el entrenamiento incrementa la capacidad oxidativa muscular (Rowell 1974). Es probable que haya mecanismos similares que aumenten el estado físico en los pacientes con TCE, aunque lo anterior no se ha analizado empíricamente.

La participación en el entrenamiento físico regular no sólo puede mejorar el estado cardiorrespiratorio (y de ese modo reducir el riesgo de morbilidad y mortalidad), sino que también puede mejorar específicamente algunas de las secuelas físicas y psicosociales comunes del TCE. Las mismas incluyen la fatiga (Bushnik 2008; Ponsford 2014), el deterioro de la cognición (Novack 2000), la depresión (Rosenthal 1998) y la integración deficiente en la comunidad (Doig 2001). La fatiga física se asocia con deterioros en la eficiencia fisiológica, la fuerza muscular y la capacidad oxidativa (Jankowski 1990). La mejoría del estado físico (es decir, la mejoría de la capacidad oxidativa) aumenta la capacidad de reserva (capacidad máxima de oxígeno menos capacidad de oxígeno en reposo); por lo tanto, para la misma actividad determinada un paciente en mejor estado físico estará trabajando a un porcentaje inferior de su capacidad máxima de oxígeno y, como tal, informará que la actividad causa menos fatiga.

Las deficiencias cognitivas después de un TCE son características, con un patrón de déficit en el funcionamiento de la memoria, la velocidad de procesamiento cognitivo, las idoneidades de razonamiento y la concentración, aunque se observan diferencias individuales (Novack 2000). Estas deficiencias son el resultado directo de la lesión cerebral traumática, con interrupción de la actividad neuronal y cambios en el metabolismo oxidativo (Grealy 1999). La justificación de la posibilidad de que el entrenamiento del estado cardiorrespiratorio quizá mejore la cognición proviene de la investigación neuropsicológica y en animales. Lo anterior indica que la actividad física regular puede afectar la cognición en forma positiva mediante el aumento del flujo sanguíneo cerebral y los niveles del neurotransmisor y produciendo cambios estructurales en el cerebro (Etnier 1997). Por lo tanto, sería posible que el entrenamiento del estado cardiorrespiratorio mejore la función cognitiva en los pacientes con TCE.

La depresión es uno de los trastornos emocionales graves que pueden seguir al TCE moderado a grave (Satz 1998); se cree que es una barrera primaria para alcanzar las metas óptimas de la rehabilitación (Rosenthal 1998). La incidencia de depresión informada después de un TCE oscila entre el 10% y el 77% (Rosenthal 1998). El mecanismo para la aparición de la depresión ha sido difícil de determinar, debido a la diversidad de las secuelas observadas después de un TCE y, en algunos casos, a los factores premórbidos que la complican, como el desempleo y el abuso de sustancias (Rosenthal 1998). Se ha postulado que el entrenamiento físico puede ser un complemento poco utilizado de los tratamientos farmacológicas y psicológicos actuales para la depresión (Paluska 2000). El entrenamiento del estado cardiorrespiratorio puede tener efectos psicológicos, como la respuesta positiva de otras personas, el aumento de la autoestima, el dominio de una nueva habilidad y el contacto social (Lawlor 2001). También pueden contribuir mecanismos fisiológicos, como el aumento que provoca el ejercicio en la liberación de endorfinas y monoaminas, que aumentan la sensación de bienestar. Probablemente, la mejor explicación del posible mecanismo del entrenamiento físico en la depresión consiste en una combinación de los componentes psicológico y fisiológico en un modelo psicobiológico (Paluska 2000).

La “integración a la comunidad?? se refiere al rendimiento efectivo de una función en el contexto comunitario (Burleigh 1998), y abarca la integración en tres áreas, a saber, la red social, el hogar y las actividades productivas (Willer 1993). En un estudio sobre supervivientes de TCE de dos a cinco años después del alta de la rehabilitación, el 40% todavía mantenía un mal funcionamiento en las tres áreas de integración en la comunidad (Doig 2001). El entrenamiento del estado cardiorrespiratorio puede cumplir una función importante para mejorar la integración en la comunidad. En primer lugar, puede repercutir en la reintegración social y familiar, pues mejora la capacidad de la persona para realizar actividad física regular con la familia o los amigos, como caminar o correr, y participar en general en las actividades socialmente aceptables independientemente de sus otros déficits (Sullivan 1990). En segundo lugar, al reducir los niveles de fatiga y mejorar el estado cardiorrespiratorio, el paciente con TCE puede ser más capaz de regresar a sus actividades cotidianas, incluidas las actividades profesionales (Jankowski 1990).

Por qué es importante realizar esta revisión

Varios estudios controlados aleatorios han examinado los efectos del entrenamiento del estado cardiorrespiratorio como una intervención para los pacientes que han sufrido un TCE (p. ej., Bateman 2001; Driver 2004), sin embargo, esta intervención no se ha sometido a revisión rigurosa. Por este motivo, se necesitó realizar una revisión sistemática para investigar integralmente estos estudios y determinar si en esta población el entrenamiento del estado cardiorrespiratorio produce beneficios o daños y para destacar las insuficiencias de la evidencia. La versión original de esta revisión se publicó en la Biblioteca Cochrane en 2008 y los resultados fueron poco claros (Hassett 2008). Debido al tiempo desde la revisión original, se consideró que una actualización era apropiada para determinar si la evidencia es más clara con respecto a la efectividad del entrenamiento físico en esta población.

Objetivos

available in

El objetivo primario de esta revisión actualizada fue evaluar si el entrenamiento mejora el estado cardiorrespiratorio en pacientes que han sufrido un TCE.

Los objetivos secundarios fueron evaluar si el entrenamiento físico mejora la función y la estructura corporal (deficiencias físicas y cognitivas, respuestas psicológicas resultantes de la lesión), las limitaciones en las actividades y las restricciones en la participación en los pacientes que han sufrido un TCE así como evaluar su seguridad, aceptación, viabilidad e idoneidad.

Métodos

available in

Criterios de inclusión de estudios para esta revisión

Tipos de estudios

Estudios controlados aleatorios que abordaron una de las dos siguientes comparaciones:

Comparación 1: cualquier programa de ejercicio (según se describe más adelante en el apartado de "Tipos de intervenciones") prescrito solo versus la atención habitual, una intervención sin ejercicios o ninguna intervención.
Comparación 2: cualquier programa de ejercicio (según se describe más adelante en el apartado de "Tipos de intervenciones") como componente de una intervención versus la atención habitual o ninguna intervención.

Tipos de participantes

Se incluyeron estudios con participantes de cualquier sexo y cualquier edad que habían sufrido un TCE (trastorno causado por un golpe o fuerza externa suficientemente importante para provocar efectos perjudiciales cerebrales [NIH 1998]), con o sin fractura del cráneo, diagnosticada por un médico con o sin confirmación mediante pruebas radiológicas. Los participantes podían tener lesiones de cualquier gravedad; cuando hubo datos disponibles, se indicó la gravedad de la lesión de los participantes en cada estudio. La gravedad de la lesión se basó en la duración de la amnesia postraumática (Russell 1961) y el coma medido según la Glasgow Coma Scale (Jennett 1981). Las tres categorías de gravedad fueron:

leve: amnesia postraumática de menos de una hora, con o sin pérdida de la conciencia, o amnesia del evento;
moderada: amnesia postraumática entre 1 y 24 horas, con o sin alteración de la conciencia (puntuación de la Glasgow Outcome Scale de 9 a 12), entre una y seis horas después del traumatismo; y
grave: duración de la amnesia postraumática de más de 24 horas, realización de un procedimiento neuroquirúrgico, o periodo de coma con una puntuación de la Glasgow Coma Scale de 8 o menos al cabo de seis horas del traumatismo.

Cuando un estudio incluía una muestra mixta de participantes con TCE y participantes con otro trastorno de salud, como un accidente cerebrovascular, sólo se usaron los resultados de los participantes con TCE. Cuando los datos no eran identificables por separado, se estableció contacto con los investigadores para obtener los datos de los pacientes con TCE solamente.

Tipos de intervenciones

Para esta revisión, se incluyeron estudios de cualquier intervención que incorporara el ejercicio de los grupos de músculos largos de forma rítmica y continua, prescrita según una frecuencia de tres veces por semana como mínimo, con una duración de al menos 20 minutos (de forma continua o intermitente) y con una intensidad que representara un claro intento de mejorar el estado cardiorrespiratorio (Palmer‐McLean 2009). Aunque no se excluyeron los estudios que no informaron la intensidad del ejercicio de al menos una captación máxima de oxígeno del 40% o una tasa del esfuerzo percibido de 13/20, tenía que estar claro que el objetivo del programa de ejercicios era el entrenamiento del estado cardiorrespiratorio. Si esta información no se aclaraba en las publicaciones de los estudios, se estableció contacto con los investigadores para conocer los parámetros del entrenamiento del programa de ejercicios.

El ejercicio se ejecutó mediante un programa para enfermos hospitalizados, pacientes ambulatorios o fuera del ámbito hospitalario, se supervisó o se autosupervisó, y se realizó en el hospital, la unidad de rehabilitación, el domicilio o la comunidad. Se incluyeron estudios con otros tipos de ejercicio, como de resistencia, flexibilidad o neuromotores, conjuntamente con el entrenamiento del estado cardiorrespiratorio. El programa de ejercicios se prescribió aislado o como un componente de una estrategia de rehabilitación.

Para que los estudios se incluyeran en la revisión, la intervención de comparación tenía que ser la atención habitual, una intervención sin ejercicios o ninguna intervención. La atención habitual podía incluir rehabilitación física, sin embargo, no podía incluir entrenamiento físico prescrito a una dosis similar a la intervención de ejercicios. En cuanto a la intervención de ejercicios, las intervenciones de comparación podían ejecutarse como un programa para enfermos hospitalizados, pacientes ambulatorios o fuera del ámbito hospitalario, supervisado o autosupervisado, y realizado en el hospital, la unidad de rehabilitación, el domicilio o la comunidad.

Tipos de medida de resultado

Los tipos de medidas de resultado usadas en cada estudio no formaron parte de los criterios de inclusión para esta revisión. En cada comparación, los análisis primarios incluyeron a todos los estudios identificados para cada resultado, tanto al final del período de intervención como al final del seguimiento programado.

Resultados primarios

El resultado primario fue el estado cardiorrespiratorio. Se evaluó mediante medidas directas e indirectas; por ejemplo:

medidas obtenidas en el laboratorio sobre las pruebas máximas, como la captación de oxígeno máxima (directa) y la potencia de salida máxima (indirecta). Otras medidas, como la frecuencia cardíaca máxima, el volumen sistólico y el cociente de intercambio respiratorio máximo durante el ejercicio, se usaron para evaluar si los rendimientos fueron máximos (ACSM 2014);
medidas de laboratorio para las pruebas submáximas, como la captación de oxígeno, la producción de dióxido de carbono, la ventilación por minuto, el cociente de intercambio respiratorio (directo) y el dióxido de carbono máximo, la duración de la prueba, y la frecuencia cardíaca y el volumen sistólico (indirecto) en condiciones de trabajo submáximas (ACSM 2014); y
medidas basadas en los participantes (indirectas), como la Rating of Perceived Exertion (RPE) (Borg 1982) y la distancia o nivel logrado en las pruebas de caminata/carrera, como la versión modificada de la prueba en la cinta rodante de 20 metros (Vitale 1997) o la prueba de 12 minutos de caminata/carrera (Cooper 1968).

La fiabilidad de la mayoría de estas medidas se ha establecido en una población con TCE (Bhambhani 2003; Mossberg 2003; Mossberg 2005; Rossi 1996; Vitale 1997).

Se calculó el tamaño de información requerido para detectar un efecto del tratamiento utilizando confiablemente un análisis del poder estadístico para un único ensayo (Bateman 2001). Si se supone una diferencia mínima clínicamente importante de 4 W para la potencia de salida máxima y una desviación estándar de 56 W, 148 pacientes deben ser asignados al azar para recibir entrenamiento físico o control con objeto de obtener un cálculo fiable del efecto del tratamiento (nivel de significación = 0,05; poder = 90%).

Resultados secundarios

Los resultados secundarios incluyeron medidas de la función y la estructura corporal (seis resultados), limitación en las actividades (dos resultados), restricción de la participación (tres resultados), seguridad, aceptación, viabilidad e idoneidad.

Medidas de la función y estructura corporal

Medidas de limitación de la actividad

Las medidas globales de la función evaluadas con el uso de cualquier escala pertinente, como el Barthel Index (Mahoney 1965), la Functional Independence Measure (Keith 1987) o la Brain Injury Community Rehabilitation Outcome Scale (Powell 1998).
La función neuromotora evaluada con el uso de cualquier medida pertinente, como la caminata interior de diez metros (van Loo 10) o cualquier escala pertinente, como la Berg Balance Scale (Berg 2003), la Motor Assessment Scale (Carr 1985), la High Level Mobility Assessment Tool (Williams 2005), el Rivermead Mobility Index (Collen 1991) o la prueba de caminata en 6 minutos (Butland 1982).

Medidas de restricción de la participación

La integración a la comunidad evaluada con el uso de cualquier medida pertinente, como las tasas de retorno al trabajo o cualquier escala pertinente, como la Sydney Psychosocial Reintegration Scale (Tate 1999), el Community Integration Questionnaire (Willer 1993) o la Craig Handicap Assessment Reporting Technique (Whiteneck 1992).
Las medidas globales de calidad de vida evaluadas con el uso de cualquier escala pertinente, como el General Health Questionnaire (Goldberg 1988), el Health Promoting Lifestyle Profile II (Walker 1995), el Bigelow Quality of Life Questionnaire (Bigelow 1991), el Sickness Impact Profile (Temkin 1988) o la Satisfaction with Life Scale (Diener 1985).
Actividad física evaluada mediante el uso de cualquier medida relevante como el número de actividades físicas informadas por el paciente o cualquier escala relevante como el International Physical Activity Questionnaire o el Active Australia (Brown 2004), o el Physical Activity Scale for Individuals with a Physical Disability (Washburn 2002).

Seguridad

Se clasificaron los efectos adversos como caídas con lesión, otras lesiones, eventos cardiovasculares importantes y cualquier otro resultado adverso.

Aceptación

Se utilizaron las tasas de abandonos para cada ensayo como una medida substitutiva de la aceptación de la intervención.

Viabilidad

La adherencia se definió como el número de sesiones de ejercicios completadas en comparación con el número prescrito de sesiones de ejercicios.

Idoneidad

La idoneidad de la intervención se definió como la capacidad de lograr los parámetros de entrenamiento prescritos (es decir intensidad, frecuencia, duración).

Results

Description of studies

Results of the search

The full details of the search results are presented in the PRISMA flow diagram (Figure 1). Six studies were included from the previous version of the review. The search strategy for this update yielded 10,711 records. Of those, 10,677 were ineligible; 28 records describing 26 studies were excluded and added to the previous nine studies described in Characteristics of excluded studies; one, NCT02276079, is ongoing and is described in Characteristics of ongoing studies; and one is awaiting assessment to confirm eligibility, McClanachan 2013, and is described in Characteristics of studies awaiting classification. Four records describing two new studies were eligible for inclusion (Driver 2009; Hoffman 2010), and were added to the six studies included in our 2008 review (Hassett 2008). A total of eight studies were included in this review update and described in Characteristics of included studies (Bateman 2001; Driver 2004; Driver 2006; Driver 2009; Hoffman 2010; Laskin 2001; McMillan 2002; Salazar 2000). Additional information about all eight studies was obtained from correspondence with the trialists and incorporated in Characteristics of included studies.

Figure 1

Study flow diagram

Included studies

All eight included studies used true randomised, parallel group design. Six studies used a two‐group design, one study randomised to three groups (McMillan 2002), and one study randomised to four groups (Laskin 2001). For the McMillan 2002 study, participants were randomised to an attentional training group, a cardiorespiratory fitness training group or a no intervention group. For analysis we compared the cardiorespiratory fitness training group to the no intervention group. For the Laskin 2001 study, participants were randomised to one of three intervention groups (cardiorespiratory fitness training alone, combined cardiorespiratory fitness and resistance training, resistance training alone) or a control group (no intervention). For analysis we combined the first two intervention groups as the experimental group, and compared this to the control group.

The eight included studies populated the following comparisons:

Comparison 1: seven studies (279 participants) compared an exercise programme prescribed alone versus a non‐exercise intervention (Bateman 2001 (47 participants); Driver 2004 (16 participants); Driver 2006 (18 participants); Driver 2009 (16 participants)), or no intervention (Hoffman 2010 (80 participants); Laskin 2001 (7 participants); McMillan 2002 (95 participants));

Comparison 2: one study (120 participants) compared an exercise programme as a component of an intervention versus usual care (Salazar 2000).

Five studies were conducted in the USA (Driver 2004; Driver 2006; Driver 2009; Hoffman 2010; Salazar 2000), two in the UK (Bateman 2001; McMillan 2002), and one in Canada (Laskin 2001). The experimental intervention was implemented in an inpatient setting for two studies (Bateman 2001; Salazar 2000), in an outpatient setting at a rehabilitation centre for one study (Laskin 2001), in a community setting for three studies (Driver 2004; Driver 2006; Driver 2009), in the home for one study (McMillan 2002), and a combination of community and home settings for one study (Hoffman 2010).

Participants

The eight included studies provide data for a total of 399 participants with TBI. Six studies included only TBI participants (Driver 2004; Driver 2006; Driver 2009; Hoffman 2010; McMillan 2002; Salazar 2000). One study included participants with a single brain injury of any cause (Bateman 2001), and another study included participants with any physical disability who attended the study centre (Laskin 2001). For these two studies, we were able to obtain the data for the TBI participants from the trialists and only these data were included in the review (Bateman 2001 for 47/175 participants; Laskin 2001 for 7/37 participants).

The participants were mainly young adult men who had sustained severe TBIs in road traffic accidents. No studies recruited children. We were unable to determine demographic or injury‐related data for the TBI‐only participants in one study (Laskin 2001). For the remaining seven studies, the average age of participants was 35 years and most were male (71%). However, all studies included women and one study included more women than men (Hoffman 2010). Injury severity was classified as severe for most participants (58%); the remainder were classified as either moderate (13%), mild (3%) or indeterminate (26%). Road traffic accident was the most common cause of TBI (41%), followed by acts of violence (13%), falls (8%), sport (1%), other (3%), or unknown causes (35%). The average time from injury to entry into the study ranged from 1.3 months in Salazar 2000 to 40.8 months for Driver 2006. Three studies recruited participants within the first year of sustaining their injury (Bateman 2001; McMillan 2002; Salazar 2000), three studies recruited participants after the first year of their injury (Driver 2004; Driver 2006; Driver 2009), and one study recruited participants between six months and five years after their injury (Hoffman 2010). Almost half (46%) of participants were ambulatory and 9% were non‐ambulatory; the ambulatory status could not be determined for 45% of participants. Because only three studies reported use of walking aids, we did not differentiate between walking unaided and walking with an aid.

Interventions

A summary of the experimental interventions (i.e. exercise programmes) is detailed in Table 1. Three studies led by the same author provided the same intervention (a community‐based aquatic exercise programme) across all three studies (Driver 2004; Driver 2006; Driver 2009). Six studies (184 participants) had fitness training as the primary focus of the study (Bateman 2001; Driver 2004; Driver 2006; Driver 2009; Hoffman 2010; Laskin 2001). The other two studies (215 participants) had cognitive training as their primary focus and included fitness training as a control group (McMillan 2002), or as a component of their experimental intervention (Salazar 2000).

See: Summary of findings for the main comparison Cardiorespiratory exercise programmes prescribed alone compared with a non‐exercise intervention or no intervention for people after traumatic brain injury

Table 1. Summary of the prescribed exercise programmes

Study	Setting	Supervised	Exercise type	Resistance exercises	Intensity	Duration	Frequency	Length of programme
Bateman 2001	Inpatient	Yes	Cycle ergometer	No	60% to 80% predicted heart rate maximum (210 − 0.65 x age)	Aim 30 minutes; average time 24.5 minutes	Aim 3 x week, average 2.1 x week	12 weeks
Driver 2004 Driver 2006 Driver 2009	Community	Yes	Aquatic	Yes	50% to 70% predicted heart rate maximum (Karvonen method)	60 minutes	3 x week	8 weeks
Hoffman 2010	1 x week community; ≥ 4 x week at home	Yes for 10 sessions, independent for 40 sessions	Participant choice (e.g. treadmill, cycle ergometer, indoor track, rowing machine, stair‐stepper)	No	60% heart rate max (220 − age) or 12/20 RPE	15 minutes education, 30 minutes warm‐up and cool‐down, 30 minutes aerobic	1 x supervised; ≥ 4 x independent	10 weeks
Laskin 2001	Outpatient	Yes	Cycle (or arm crank) ergometer	Yes	Heart rate and power output set to second ventilatory threshold from peak test	Average time 15 minutes week 1 to 28.5 minutes week 12 for continuous fitness component	3 x week	12 weeks
McMillan 2002	Home	Yes for 5 sessions, independent for 15 sessions	Basic fitness exercises, no equipment	Unknown	Not specified	45 minutes	5 x week	4 weeks
Salazar 2000	Inpatient	Yes	Fitness group	Unknown	At own pace	60 minutes	5 x week	8 weeks

Abbreviation RPE = rated perceived exertion

The length of the exercise programme was four weeks (McMillan 2002), eight weeks (Driver 2004; Driver 2006; Driver 2009; Salazar 2000), 10 weeks (Hoffman 2010), or 12 weeks (Bateman 2001; Laskin 2001). Supervision of the exercise programme was provided for six of the eight studies (Bateman 2001; Driver 2004; Driver 2006; Driver 2009; Laskin 2001; Salazar 2000); the other two studies had a mix of supervised and independent sessions over the duration of the programme (Hoffman 2010; McMillan 2002). The exercise programme included a combination of cardiorespiratory fitness and resistance training for four studies (Driver 2004; Driver 2006; Driver 2009; Laskin 2001), cardiorespiratory fitness training alone for two studies (Bateman 2001; Hoffman 2010), and was unclear for two studies (McMillan 2002; Salazar 2000). The exercise programme was compared to a non‐exercise intervention in four studies (Bateman 2001; Driver 2004; Driver 2006; Driver 2009), no intervention in two studies (Laskin 2001; McMillan 2002), a usual care intervention in one study (Salazar 2000), and a waiting‐list control group who were offered the programme at the end of the wait‐list period (Hoffman 2010).

The six studies with fitness training as their primary focus prescribed the cardiorespiratory fitness component of the exercise programme (type, intensity, duration, frequency) within the ACSM guidelines for people who have sustained a brain injury (Palmer‐McLean 2009). Only one of the six studies reported the actual exercise dose delivered (Bateman 2001), which was below the ACSM recommendations (Palmer‐McLean 2009). In the two studies for which fitness training was not the primary focus, it was not possible to determine if the exercise programme met the ACSM guidelines; however, it was clear that the intention of the intervention was cardiorespiratory fitness training (McMillan 2002; Salazar 2000). Training load was progressed in one study by increasing the duration and power output (Laskin 2001), in another study by increasing the duration (Bateman 2001), and was not reported for the other six studies. From correspondence with one trialist, the exercise programme focused on reaching a perceived exertion level and the training load was progressed by the participant by increasing the difficulty as their exertion level allowed (Hoffman 2010).

Outcomes

Only the immediate effects of intervention were examined for five studies (Driver 2004; Driver 2006; Driver 2009; Hoffman 2010; Laskin 2001), with outcomes being measured at baseline and at the end of intervention. A follow‐up period between 12 weeks and 12 months was used to determine the sustained effects of the intervention in two studies (Bateman 2001; McMillan 2002). Only long‐term effects were examined for one study, with the trial's primary outcome measures (i.e. return to employment and fitness for duty) being reported at one year follow‐up only (Salazar 2000).

The primary outcome of cardiorespiratory fitness was measured in three studies (Bateman 2001; Driver 2004; Laskin 2001). The eight included studies measured a mix of outcomes in the body function and structure, activity limitation and participation restriction domains, but the outcomes measured varied considerably among studies. At the body function and structure level, the secondary outcome of body composition was measured in two studies (Bateman 2001; Driver 2004), strength in two studies (Driver 2004; Laskin 2001), fatigue in three studies (Bateman 2001; Driver 2009; Hoffman 2010), cognition in three studies (Driver 2009; McMillan 2002; Salazar 2000), and depression in five studies (Bateman 2001; Driver 2009; Hoffman 2010; McMillan 2002; Salazar 2000). For activity limitation, the secondary outcome measure of global function was measured in one study (Bateman 2001), and neuromotor function was measured in two studies (Bateman 2001; Hoffman 2010). For participation restriction, the secondary outcome of community integration was measured in two studies (Hoffman 2010; Salazar 2000), and quality of life in three studies (Driver 2006; Hoffman 2010; McMillan 2002). The secondary outcome measures of motivation and physical activity were not measured in any of the eight included studies.

Safety (adverse events) was not reported in the eight studies. Trialists were contacted and adverse events data were provided for six studies (Bateman 2001; Driver 2004; Driver 2006; Hoffman 2010; Laskin 2001; Salazar 2000). Acceptance data (i.e. dropout data) were available for all studies. Feasibility data (adherence to the prescribed exercise programme) were available for six studies (Bateman 2001; Driver 2004; Driver 2006; Driver 2009; Hoffman 2010; Laskin 2001). Only one study provided data to determine the suitability of fitness training in people with TBI (Bateman 2001).

Excluded studies

We excluded 35 studies for the following reasons:

10 did not fulfil our Types of studies criteria (Corral 2014; Dault 2002; Dresen 1985; Driver 2003; Driver 2016; Esquenazi 2013; Esquenazi 2017; Freivogel 2009; Hassett 2009; Kim 2014);
six did not fulfil our Types of participants criteria (Calder 2008; Elsworth 2011; McAuley 2001; Rarick 1968; Sherrington 2008; Varela 2012);
15 did not fulfil our Types of interventions criteria (Baque 2017; Bellon 2015; Bender 2016; Berrol 1997; Blake 2009; Canning 2003; Clanchy 2016; Hassett 2012; Haworth 2009; Katz‐Leurer 2009; Kurowski 2017; Maerlender 2015; Shiel 2001; Slade 2002; Zhu 2007);
three trials provided too little information, even after correspondence with authors, for us to be certain that they met our inclusion criteria (Froehlich‐Grobe 2004; Herterich 2002; Zhu 2004); and
one trial was terminated due to recruitment issues (Reed 2015).

Further details of these trials are provided in Characteristics of excluded studies.

Risk of bias in included studies

The risk of bias is reported for each study following each Characteristics of included studies table and is summarised across all the included studies in Figure 2 and for each included study in Figure 3. We required clarification from trialists for seven of the eight studies for at least one of the eight risk of bias domains to confirm our judgements (Driver 2004; Driver 2006; Driver 2009; Hoffman 2010; Laskin 2001; McMillan 2002; Salazar 2000). All studies had high and low risk of bias judgements for the different domains, and five studies had at least one domain that was rated at unclear risk of bias (Bateman 2001; Driver 2009; Hoffman 2010; McMillan 2002; Salazar 2000).

Figure 2

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

Figure 3

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

Allocation

All studies randomised participants to study groups, however there was uncertainty regarding the method used to generate the random sequence for two studies (McMillan 2002; Salazar 2000). Four studies reported using allocation concealment when assigning participants to study groups (Bateman 2001; Hoffman 2010; Laskin 2001; Salazar 2000), three studies did not use allocation concealment (Driver 2004; Driver 2006; McMillan 2002), and allocation was unclear for one study (Driver 2009).

Blinding

Blinding of participants and personnel delivering the intervention (performance bias) is not possible in exercise studies; therefore, we judged this domain to be at high risk of bias for all studies. Detection bias was split into objective and self‐report measures. For the outcomes included in this review, six studies included objective measures (Bateman 2001; Driver 2004; Hoffman 2010; Laskin 2001; McMillan 2002; Salazar 2000), and five studies included self‐reported measures (Bateman 2001; Driver 2006; Driver 2009; Hoffman 2010; McMillan 2002). Four of the six studies that measured objective outcomes used blinded assessors to measure the key outcomes and we judged these to be at low risk of detection bias (Bateman 2001; Driver 2004; Hoffman 2010; McMillan 2002). One of the remaining two studies that measured objective outcomes did not use a blinded assessor and was assessed at high risk of detection bias (Laskin 2001). Salazar 2000 was not able to use blinded assessors for the primary outcome measure of return to work (this outcome was determined independently of the study and in accordance with army regulations) and was judged to have a low risk of detection bias. Three of the five studies that measured self‐reported outcomes used blinded assessors to collect these data (Bateman 2001; Hoffman 2010; McMillan 2002); however, because participants were not blinded to group allocation, detection bias was rated as unclear. We rated detection bias as high for the two studies that did not use a blinded assessor and where the participants were unblinded to group allocation (Driver 2006; Driver 2009).

Incomplete outcome data

Seven of the eight included studies measured their outcomes at the end of intervention. Four of these studies had no dropouts and complete data at the end of intervention for primary outcome measures (Driver 2004; Driver 2006; Driver 2009; Laskin 2001); we therefore rated these studies at low risk of attrition bias. Two of the remaining three studies had a small proportion of dropouts and the completeness of data at the end of intervention for the primary outcome measure ranged from 90% to 94% (Bateman 2001; Hoffman 2010). We also rated these studies at low risk of attrition bias. One study had a 19% dropout rate in the fitness training group (but did not report the reasons) and no dropouts in the control group (McMillan 2002); we rated this trial at high risk of attrition bias.

Three of the included studies conducted follow‐up assessments. The completeness of data for the primary outcome measure at follow‐up ranged from 77% to 100% (Bateman 2001; McMillan 2002; Salazar 2000). Although Salazar 2000 reported 89% follow‐up for the trial's primary outcomes (return to work and fitness for duty), only 63% of participants were followed up in person to conduct the cognitive outcome measures; we therefore rated this study at high risk of attrition bias.

Selective reporting

None of the included studies was registered on clinical trial registries, so we could not confirm that all outcome measures were reported. We therefore rated all studies at high risk of reporting bias.

Other potential sources of bias

We were unable to identify any other potential sources of bias for seven of the eight studies. For one study (Salazar 2000), bias may have been introduced from potential contamination of the control group in which participants were instructed to resume their usual physical activity. This study's participants came from the armed forces and so usual physical activity for some may have included exercising at the same or an even higher dosage of exercise than the experimental group. We rated this study as being at an unclear risk of bias for this domain.

Effects of interventions

See: summary of findings Table for the main comparison for the main comparison, Comparison 1: cardiorespiratory exercise programmes prescribed alone compared with a non‐exercise intervention or no intervention for people after TBI.

We pooled data from the end of the intervention for seven outcomes within Comparison 1 (cardiorespiratory fitness, body composition, strength, fatigue, depression, neuromotor function and dropouts) and pooled data for a post hoc analysis for a eighth outcome (depression). For five of the eight pooled outcomes (body composition, strength, fatigue, depression and post hoc depression), different studies used different outcome measures. In these instances, we calculated the standardised mean difference (SMD) and 95% confidence intervals (CI). For each of the pooled outcomes we applied the GRADE criteria to rate the quality of the evidence and the justification for each rating is provided in Table 2. The remaining outcomes were not pooled due to a large number of single study outcomes, due to substantial heterogeneity, or due to no events being reported (one outcome). It is likely that heterogeneity existed due to both clinical and methodological diversity among studies.

Table 2. Justification of GRADE ratings for pooled outcomes

Outcome	Reasons to downgrade the evidence					Reasons to upgrade the evidence	GRADE rating
Outcome	Risk of bias of study design	Indirectness of evidence	Heterogeneity	Imprecision	Publication bias	Large magnitude of effect, dose response, effect of all plausible confounding factors	GRADE rating
Cardiorespiratory fitness (3 studies, 67 participants)	Downgraded x 1 level: allocation concealment at high risk of bias for 1 study; no blinded assessor for 1 study	Nil: not gold standard measure, but still common measure of fitness	Nil: I² = 36%; Chi² P = 0.21	Downgraded x 1 level: wide CI and 2 studies with positive and 1 study with negative mean differences	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 2 levels to LOW
Body composition (2 studies, 61 participants)	Downgraded x 1 level: allocation concealment high for 1 of 2 studies	Nil	Nil: I² = 0% Chi² P = 0.88	Downgraded x 1 level: wide CI	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 2 levels to LOW
Strength (2 studies, 23 participants)	Downgraded x 2 levels: allocation concealment high for 1 of 2 studies; unblinded assessor for 1 of 2 studies; small number of participants	Nil	Nil: I² = 0% Chi² P = 0.77	Downgraded x 1 level: wide CI and 1 study with positive and 1 study with negative mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 3 levels to VERY LOW
Fatigue (3 studies, 130 participants)	Downgraded x 1 level: allocation concealment unclear for 1 study and low for 2 studies; high risk (1 study) or unclear risk (2 studies) of detection bias: participants not blinded for self‐reported measure	Nil	Downgraded x 1 level: I² = 56% Chi² P = 0.11	Downgraded x 1: wide CI and 2 studies with negative and 1 study with positive mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 3 levels to VERY LOW
Depression (4 studies, 220 participants)	Downgraded x 1 level: allocation concealment high for 1 of 4 studies and unclear for 1 of 4 studies; high risk (1 study) or unclear risk (3 studies) of detection bias: participants not blinded for self‐reported measure	Nil	Downgraded x 1 level: I² = 64% Chi² P = 0.04	Downgraded x 1: wide CI and 3 studies with negative and 1 study with positive mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 3 levels to VERY LOW
Depression post hoc analysis (3 studies, 178 participants)	Downgraded x 1 level: allocation concealment high for 1 of 4 studies and unclear for 1 of 4 studies; high risk (1 study) or unclear risk (3 studies) of detection bias: participants not blinded for self‐reported measure	Nil	Downgraded x 1 level: I² = 53% Chi² P = 0.04	Nil	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 2 levels to LOW
Neuromotor function (2 studies, 109 participants)	Nil	Nil	Nil: I² = 28% Chi² P = 0.24	Downgraded x 1 level: wide CI and 1 study with positive and 1 study with negative mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 1 level to MODERATE
Acceptance ‐ dropout rate (3 studies, 222 participants)	Downgraded x 1 level: allocation concealment high for 1 of 3 studies	Nil	Downgraded x 1 level: I² = 50% Chi² P = 0.13	Downgraded x 1 level: wide CI and 2 studies with negative and 1 study with positive mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 3 levels to VERY LOW

Abbreviation CI = confidence interval

Comparison 1: exercise programme prescribed alone versus usual care, non‐exercise intervention or no intervention

Cardiorespiratory fitness (primary outcome)

Only one study assessed cardiorespiratory fitness using peak oxygen uptake at the endpoint of an incremental cycle ergometer test, however, this study included only seven TBI participants (Laskin 2001). Three of the seven studies assessed cardiorespiratory fitness indirectly at the end of intervention using power output at the endpoint of an incremental cycle ergometer test; these data were pooled using a random‐effects model (Bateman 2001; Driver 2004; Laskin 2001; 67 participants; Analysis 1.1). One study also measured power output at follow‐up (Bateman 2001; Analysis 1.1). We downgraded the quality of the evidence for this outcome to low due to risk of bias of the included studies and imprecision of the estimate of effect. There was a mean improvement of 35.47 W on an exercise test in the fitness training intervention compared to the control intervention, however this estimate was imprecise with 95% CI 2.53 to 68.41 W. The mean improvement of 35 W represents approximately a 36% improvement in fitness from baseline measurement (average baseline peak power output across the three studies = 97 W). Although a minimal clinically important difference (MCID) has not been determined for this measure in this or similar populations, in people with severe chronic obstructive pulmonary disease ‐ whose baseline power output is approximately one‐third of the baseline power output in the included studies ‐ MCID has been determined to be 4 W (Puhan 2011). Taking this into consideration, the confidence intervals are likely to include both a substantial clinically important effect and a possible negligible effect. There was no clear long‐term effect of fitness training on cardiorespiratory fitness (MD 3.27 W, 95% CI ‐34.21 to 40.75; 1 study, 40 participants; Analysis 1.1).

Body function and structure (impairment): body composition (secondary outcome)

Two studies measured body composition at the end of the intervention (Bateman 2001; Driver 2004); Bateman 2001 also measured body composition at the end of follow‐up. We pooled the immediate effect of intervention using a random‐effects model. The effect size was small and in favour of the control intervention, however the wide confidence intervals (suggesting a positive or negative effect) mean we are uncertain about the effects of fitness training on body composition (SMD 0.29, 95% CI ‐0.22 to 0.79; 2 studies, 61 participants; low‐quality evidence; Analysis 1.2). No clear long‐term effect of fitness training on body composition was evident (SMD 0.52, 95% CI ‐0.11 to 1.14; 1 study, 41 participants; Analysis 1.2).

Body function and structure (impairment): strength (secondary outcome)

Two studies measured strength at the end of intervention (Driver 2004; Laskin 2001). We pooled the immediate effect of intervention using a random‐effects model. The between‐group difference was negligible and the confidence intervals indicate considerable uncertainty about whether fitness training is effective at increasing muscle strength (SMD ‐0.02, 95% CI ‐0.86 to 0.83; 2 studies, 23 participants; very low‐quality evidence; Analysis 1.3).

Body function and structure (impairment): fatigue (secondary outcome)

Three studies measured fatigue at the end of intervention (Bateman 2001; Driver 2009; Hoffman 2010); Bateman 2001 also measured fatigue at end of follow‐up. The three studies used different measures of fatigue including a visual analogue scale (VAS) (Hoffman 2010), the fatigue subscale of the Profile of Moods State (Driver 2009), and the modified version of the Chalder Fatigue Scale (Bateman 2001). The Chalder Fatigue Scale provides both a physical and mental fatigue score; we pooled the physical subscale score with the measures from the other two studies because these were more in the physical fatigue domain. We downgraded the quality of the evidence for this outcome to very low due to risk of bias, heterogeneity and imprecision. There was a small reduction in self‐reported fatigue with fitness training compared to the control interventions, however the wide confidence intervals indicated imprecision for this estimate and a potentially positive or negative effect of fitness training on fatigue (SMD ‐0.32, 95% CI ‐0.90 to 0.26; 3 studies, 130 participants; very low‐quality evidence; Analysis 1.4). We transformed this SMD into a score on the VAS‐fatigue scale (score 0 to 10, using the average baseline score from Hoffman 2010), which showed this result would equate to an average reduction in fatigue of 0.9 points, with the 95% confidence intervals indicating an effect between a 2.6 point reduction in fatigue to a 0.7 point increase in fatigue with fitness training compared to the control interventions. There was no clear long‐term effect of fitness training on physical fatigue (SMD 0.35, 95% CI ‐0.28 to 0.97; 1 study, 40 participants; Analysis 1.4).

Body function and structure (impairment): cognition (secondary outcome)

One study measured the cognitive domains of memory and attention at the end of intervention and at the end of follow‐up (McMillan 2002; Table 3). In addition, the Profile of Moods State measured in Driver 2009 included a confusion subscale which could be classified as a measure of cognitive function. However, we felt it did not measure cognition in the same detail as the neuropsychological tests used in McMillan 2002, and so we did not pool this outcome. There were no immediate or long‐term effects of fitness training on cognitive function compared to the control interventions in either study (Table 3).

Table 3. Additional data for comparison 1

Outcome	Study	Outcome measure	Timing	Intervention group Mean (SD); N	Control group Mean (SD); N	Mean between group difference (MD), 95% CI
Cognition	McMillan 2002	PASAT	End of intervention	37 (13); 38	32 (14); 48	MD 5, 95% CI ‐0.7 to 10.7
		PASAT	End of follow‐up	38 (14); 35	34 (15); 38	MD 4, 95% CI ‐2.7 to 10.7
		Sunderland Memory Questionnaire	End of intervention	52 (35); 38	52 (37); 48	MD 0, 95% CI ‐15.3 to 15.3
		Sunderland Memory Questionnaire	End of follow‐up	49 (36); 35	47 (34); 38	MD 2, 95% CI ‐14.1 to 18.1
		Adult Memory & Information Processing Battery	End of intervention	24 (12); 38	23 (12); 48	MD 1, 95% CI ‐4.1 to 6.1
		Adult Memory & Information Processing Battery	End of follow‐up	34 (10); 35	31 (13); 38	MD 3, 95% CI ‐2.3 to 8.3
	Driver 2009	POMS: confusion subscale	End of intervention	0.6 (0.3); 8	0.9 (0.6); 8	MD ‐0.2, 95% CI‐0.7 to 0.2
Community integration	Hoffman 2010	CHART‐SF	End of intervention	85 (14); 37	83 (12); 39	MD 1.7, 95% CI ‐4.1 to 7.5
Global measure of function	Bateman 2001	Barthel Index	End of intervention	17 (4); 24	18 (3); 22	MD ‐0.9, 95% CI ‐2.9 to 1.2
Global measure of function	Bateman 2001	Barthel Index	End of follow‐up	18 (4); 22	19 (3); 21	MD ‐1.1, 95% CI ‐3.3 to 1.2

Abbreviations CHART‐SF = Craig Handicap Assessment and Reporting Technique ‐ Short Form; PASAT = Paced Auditory Serial Addition Test; POMS = Profile of Mood States

Body function and structure (psychological functioning): depression (secondary outcome)

Four studies measured depression at the end of intervention (Bateman 2001; Driver 2009; Hoffman 2010; McMillan 2002); Bateman 2001 and McMillan 2002 also measured depression at the end of follow‐up. The four studies used three different outcome measures of depression; the depression subscale of the Hospital Anxiety Depression Scale (Bateman 2001; McMillan 2002), the depression‐dejection subscale of the Profile of Moods States (Driver 2009) and the Beck Depression Inventory (Hoffman 2010). The quality of the evidence for this outcome was downgraded to very low due to risk of bias, heterogeneity and imprecision. There was a small reduction in self‐reported depression with fitness training compared to the control interventions, however the wide confidence intervals indicated imprecision of this estimate and potentially a positive or negative effect of fitness training on depression (SMD ‐0.43, 95% CI ‐0.92 to 0.06; 4 studies, 220 participants; very low‐quality evidence; Analysis 1.5). Transforming the SMD into a score on the depression subscale of the Hospital Anxiety Depression Scale (score 0 to 21 with a higher score indicating greater depression, using the average baseline score from Bateman 2001 and McMillan 2002), this result would equate to an average reduction in depression of 1.7 points with the 95% confidence intervals indicating an effect between a 3.7 point reduction in depression to a 0.2 point increase in depression with fitness training compared to the control interventions. At the end of follow‐up there was substantial heterogeneity (I² = 74%, P = 0.05), therefore we did not pool the data. There was no clear long‐term effect of fitness training on depression for either study (SMD ‐0.44, 95% CI ‐0.90 to 0.03; McMillan 2002, 73 participants; SMD 0.35, 95% CI ‐0.28 to 0.97; Bateman 2001, 40 participants; Analysis 1.5).

There was moderate heterogeneity for depression at the end of intervention (I² = 64%, P = 0.04), however this was not explained by methodological diversity (studies with low risk of bias from allocation concealment; I² = 60%, P = 0.11; Bateman 2001; Hoffman 2010). Our post hoc analysis, in which we pooled the data from the three studies that included subacute and chronic participants at home only (Driver 2009; Hoffman 2010; McMillan 2002), yielded an SMD that indicated a moderate reduction in depression with fitness training immediately after intervention, with the 95% CIs indicating that the effect lay between a large and small reduction in depression with fitness training compared to the control interventions (SMD ‐0.61, 95% CI ‐1.1 to ‐0.11; 3 studies, 178 participants; low‐quality evidence; Analysis 1.9). The fourth study, which included participants who were still participating in inpatient rehabilitation, showed a small reduction in depression in favour of the control intervention (relaxation therapy), however the 95% CIs were wide indicating uncertainty regarding which intervention was more effective at reducing depression (SMD 0.19, 95% CI ‐0.42 to 0.80, 42 participants; Bateman 2001).

Body function and structure (psychological functioning): motivation (secondary outcome)

No studies measured motivation.

Activity limitation: global (secondary outcome)

Only one study measured the effect of the intervention on a global measure of function (Bateman 2001; Table 3). There was no clear immediate or long‐term effect of fitness training on function globally.

Activity limitation: neuromotor function (secondary outcome)

Two studies measured neuromotor function at the end of intervention during a walking test in metres per second (m/s) (Bateman 2001; Hoffman 2010; Analysis 1.6). We pooled the immediate effect of fitness training on neuromotor function using a random‐effects model. There was uncertainty regarding the effect of fitness training on walking speed with the confidence intervals spanning a difference of approximately 0.3 m/s in either direction (MD 0.01 m/s, 95% CI ‐0.25 to 0.27; 2 studies, 109 participants; moderate‐quality evidence; Analysis 1.6). A minimal clinically important difference (MCID) in people with TBI was determined to be 0.15 m/s for a comfortable walking speed over 10 m (van Loo 2004).

Participation restriction: community integration (secondary outcome)

One study measured community integration at the end of intervention with no clear evidence of effect (Hoffman 2010; Table 3).

Participation restriction: quality of life (secondary outcome)

Three studies measured quality of life at the end of intervention (Driver 2006; Hoffman 2010; McMillan 2002), with only one measuring it also at end of follow‐up (McMillan 2002). The outcome measure used in the McMillan 2002 study (General Health Questionnaire) gives a higher score for a worse outcome. To match the other two studies where a higher score equals a better outcome, we subtracted the mean scores for each group in the McMillan 2002 study from the maximum possible score for this outcome measure. There was substantial heterogeneity for this outcome (I² = 93%; P < 0.01); and so we did not pool data. We could not test the effect of allocation concealment on heterogeneity, as two of the three studies did not conceal allocation, indicating considerable risk of bias for this outcome. There were mixed results among studies for this outcome. Self‐reported quality of life was rated as better in the intervention group than the control group at the end of intervention for one of the three studies (SMD 26.04, 95% CI 16.37 to 35.71; Driver 2006, 18 participants; Analysis 1.7), however this was not measured at end of follow‐up to determine if this effect was sustained. For the other two studies, it was uncertain whether quality of life was rated as better or worse at the end of the intervention for participants who completed fitness training compared to control interventions (SMD 0.42, 95% CI ‐0.03 to 0.88; 76 participants, Hoffman 2010; SMD 0.38, 95% CI ‐0.05 to 0.81; 86 participants, McMillan 2002), and at end of follow‐up (SMD 0.15, 95% CI ‐0.31 to 0.61; 73 participants, McMillan 2002).

Participation restriction: physical activity (secondary outcome)

No studies measured physical activity.

Safety

The trial publications reported no adverse events for the intervention group in any of the seven studies included in Comparison 1. Correspondence with trialists produced confirmation of no adverse events for five studies (Bateman 2001; Driver 2004; Driver 2006; Laskin 2001; McMillan 2002), and no clarification for two studies (Driver 2009; Hoffman 2010), which indicated that there was no evidence of harm of providing cardiorespiratory fitness training for participants.

Acceptance

The five studies that provided supervised fitness training sessions had no dropouts from their intervention groups at the end of intervention (Bateman 2001; Driver 2004; Driver 2006; Driver 2009; Laskin 2001). The other two studies had a mix of supervised and unsupervised sessions and had a dropout rate of 19% (9/47) for McMillan 2002 (reasons not documented) and 8% (3/40) for Hoffman 2010 (unable to contact participants) from the fitness training group. For the three studies with dropouts in either group (Bateman 2001; Hoffman 2010; McMillan 2002), we pooled the dropout rate data using a random‐effects model (222 participants, 3 studies; very low‐quality evidence; Analysis 1.8). Participants were three times less likely to drop out of the control group than the intervention group, however this odds ratio estimate was imprecise, with the 95% confidence intervals spanning a range between 0.32 in favour of the fitness training group to 30 times more in favour of the control intervention. The overall low rate of dropouts indicates that an exercise programme incorporating cardiorespiratory fitness training is an acceptable intervention for participants.

Feasibility

Training adherence was reported as 69% in Bateman 2001, 84% in Driver 2006 and Driver 2009, and 100% in Driver 2004 and Laskin 2001, and was not reported for McMillan 2002. The seventh study, Hoffman 2010, reported mean attendance of the once a week supervised session as 5.9 sessions out of 10, with more than half of the intervention group attending seven or more of the 10 supervised sessions and four attending none of the sessions. These results indicated that an exercise programme incorporating cardiorespiratory fitness training is a feasible intervention.

Suitability

Only one study reported how well the participants achieved the training parameters (Bateman 2001). In this study, fitness training was set at an intensity of 60% to 80% of age‐predicted maximal heart rate, for a duration of 30 minutes, with a frequency of three times per week, for 12 weeks (36 sessions). On average, the participants in the experimental group spent 66% (16.2 minutes out of 24.5 minutes) of exercise time within or above their training heart rate zone for an average of 2.1 sessions per week. Overall, there was insufficient evidence to determine how suitable the prescribed interventions (i.e. training parameters) were for participants.

Comparison 2: exercise programme as a component of an intervention versus usual care

Only one study with 120 participants was included in Comparison 2 (Salazar 2000). No data were available to assess the immediate effects of intervention; only the long‐term effects were reported and the data are reported in Table 4.

Table 4. Additional data for comparison 2

Outcome	Study	Outcome measure	Timing	Intervention group Mean (SD); N	Control group Mean (SD); N	Between‐group difference
Cognition	Salazar 2000	Buschke Selective Reminding Test	End of follow‐up	67 (34); 42	63 (40); 34	MD 4.0, 95% CI ‐12.9 to 20.9
Cognition		PASAT	End of follow‐up	42 (14); 42	38 (12); 33	MD 3.5, 95% CI ‐2.3 to 9.4
Depression		Diagnosis of major depression	End of follow‐up	7/42^a	9/34^a	OR 0.6, 95% CI 0.2 to 1.7
Community integration		Return to work	End of follow‐up	60/67^a	50/53^a	OR 0.5, 95% CI 0.1 to 2.1
Acceptance		Dropouts	During the intervention phase	7/67^a	2/53^a	OR 3.0, 95% CI 0.6 to 15.0

^aNumber of events over total number of cases

Abbreviations MD = mean difference; OR = odds ratio; PASAT = Paced Auditory Serial Addition Test

Cardiorespiratory fitness (primary outcome)

No data available.

Body function and structure (impairment): body composition, strength, fatigue (secondary outcomes)

No data available.

Body function and structure (impairment): cognition (secondary outcome)

Two outcome measures were used to measure cognition; the Paced Auditory Serial Addition Test (PASAT) and the Buschke Selective Reminding Test. At the end of follow‐up, it was uncertain whether cognition was rated as better or worse for participants in the intervention and control groups (Table 4).

Body function and structure (psychological functioning): depression (secondary outcome)

Twenty‐one per cent (16/76) of participants were diagnosed with major depression at the end of follow‐up, with no difference in depression between the intervention and control groups (Table 4).

Body function and structure (psychological functioning): motivation (secondary outcome)

No data available.

Activity limitation: global and neuromotor function (secondary outcomes)

No data available.

Participation restriction: community integration (secondary outcome)

Most participants had returned to work at the end of follow‐up (92%). There was no difference in return to work rates between the intervention and control groups (Table 4).

Participation restriction: quality of life and physical activity (secondary outcomes)

No data available.

Safety

No adverse events were reported for the intervention group although six were reported in the control group. These adverse events were classified in the 'any other adverse events' category (i.e. they required supplemental therapy either during or straight after the control intervention because of persisting behaviour or mood problems). The control intervention for that study was an independent home programme (including cognitive training and an exercise programme), so it is unlikely that the need for additional intervention was strongly related to the independent home exercise programme.

Acceptance

There was a 10% (7/67) dropout rate in this study, two for medical reasons and five voluntary withdrawals. There were no differences in dropout rates between the intervention and control groups (Table 4).

Feasibility

No data available.

Suitability

No data available.

Discusión

available in

Hay dos estudios nuevos incluidos en esta revisión actualizada, lo cual genera una total de ocho estudios controlados aleatorios (399 participantes) que han investigado los efectos del entrenamiento del estado cardiorrespiratorio en pacientes que han sufrido un traumatismo craneoencefálico (TCE). Los estudios identificados incorporaron intervenciones clínicamente diferentes ejecutadas en diversos momentos después de la lesión y midieron una variedad de resultados con numerosas herramientas de medición diferentes. La diversidad clínica y metodológica entre los estudios y el número de resultados de estudios únicos impidió el agrupamiento de los resultados para todos menos siete de los resultados evaluados en los estudios incluidos. Además, cuando los estudios se agruparon, la calidad de la evidencia se disminuyó de alta a moderada, baja o muy baja. En consecuencia, hay incertidumbre con respecto a los resultados, lo cual dificulta la extracción de conclusiones definitivas con respecto a la efectividad del entrenamiento del estado cardiorrespiratorio después del TCE.

Resumen de los resultados principales

Resultado primario

El objetivo primario de esta revisión actualizada fue evaluar si el entrenamiento mejora el estado cardiorrespiratorio en pacientes que han sufrido un TCE. La medida de referencia (gold standard) para el estado cardiorrespiratorio en esta población es la captación máxima de oxígeno, aunque sólo uno de los tres estudios incluidos que evaluaron el estado cardiorrespiratorio usó esta medida (Laskin 2001). Los otros dos estudios (y el estudio Laskin 2001) usaron una medida indirecta del estado cardiorrespiratorio, es decir, la potencia de salida máxima lograda durante una prueba gradual submáxima con un criterio de valoración predeterminado (es decir, porcentaje de la frecuencia cardíaca máxima teórica) (Bateman 2001; Driver 2004; Laskin 2001). La diferencia media entre los grupos para los tres estudios incluidos en este análisis fue de 35,5 W que probablemente es una diferencia clínicamente importante. Sin embargo, el intervalo de confianza del 95% varió de 2,5 a 68,4 W, lo cual destaca la imprecisión de este cálculo y que el efecto puede no ser clínicamente relevante. Sólo uno de los tres estudios mostró un efecto positivo; este estudio reclutó a participantes que estaban en la fase crónica, y que, en promedio, se realizaron después de más de tres de años desde la lesión (Driver 2004). Por el contrario, uno de los estudios que no reveló un efecto claro ejecutó el entrenamiento físico en la fase aguda de la rehabilitación durante la hospitalización (Bateman 2001), con participantes que, en promedio, habían sufrido la lesión hacía menos de seis meses. Los participantes tanto de los grupos de intervención como de control de dicho estudio todavía podrían haber estado mejorando de forma aguda, debido a que los tres a seis primeros meses luego de la lesión se consideran el período con la recuperación más rápida (Ponsford 2014). En consecuencia, esto puede haber reducido cualquier diferencia potencial entre los grupos. Se necesitan estudios adicionales para investigar el entrenamiento físico en los estadios tanto agudos como crónicos de la recuperación y para medir la captación máxima de oxígeno con objeto de determinar si el entrenamiento físico es más efectivo en momentos particulares del proceso de rehabilitación.

Medidas de resultado secundarias (deficiencias en la función y la estructura corporal, limitación de las actividades, restricción de la participación)

El segundo objetivo era investigar si el entrenamiento del estado físico mejora la función y la estructura corporal, la limitación de las actividades o la restricción de la participación en los pacientes que han sufrido un TCE. No se pudo determinar un efecto claro de la intervención sobre ninguno de los resultados secundarios, principalmente debido a la falta de datos. Dentro de los resultados secundarios, fue posible agrupar los datos sólo para cinco resultados (depresión, fatiga, fuerza, composición corporal y función neuromotora), y se agruparon los datos en un análisis post hoc para la depresión. Ningún estudio incluyó medidas de la motivación o la actividad física. Los intervalos de confianza fueron amplios para los cinco resultados agrupados, y no se conoce si el entrenamiento físico fue más o menos efectivo para mejorar el resultado en comparación con las intervenciones de control. Este hecho destaca la necesidad de realizar estudios con poder estadístico suficiente, que incorporen medidas de la función y estructura corporal, limitación de actividades y restricción de la participación, para evaluar de forma satisfactoria los beneficios del entrenamiento del estado físico de los pacientes que han padecido un TCE.

Medidas de resultado secundarias (seguridad, aceptación, factibilidad, idoneidad)

Los objetivos secundarios también procuraron evaluar la seguridad, la aceptación, la viabilidad y la idoneidad del entrenamiento físico en los pacientes con TCE. No se informaron eventos adversos en los estudios incluidos; por lo tanto, no hay evidencia de efectos perjudiciales al proporcionar entrenamiento del estado físico como una intervención para los pacientes con TCE. Sin embargo, la selección de los participantes y la supervisión durante el entrenamiento pueden haber influido en este resultado. Cinco de los ocho estudios incluidos usaron monitores de la frecuencia cardíaca y la supervisión de uno en uno durante cada sesión de entrenamiento (Bateman 2001; Driver 2004; Driver 2006; Driver 2009; Laskin 2001) y un sexto estudio proporcionó supervisión personalizada para una sesión por semana (Hoffman 2010), por lo tanto los participantes fueron vigilados para asegurar que se estaban ejercitando de forma segura y a una intensidad apropiada del ejercicio. También se debe considerar que el resultado de que no hubo eventos adversos puede deberse a que no se preguntó específicamente a los participantes, por lo que los eventos adversos menores, como un esfuerzo muscular excesivo, pueden no haberse informado. La selección de la idoneidad de los participantes potenciales para hacer ejercicios fue también realizada en seis de los ocho estudios incluidos (Bateman 2001; Driver 2004; Driver 2006; Driver 2009; Hoffman 2010; Laskin 2001), lo cual puede haber evitado los eventos adversos. Debe observarse, sin embargo, que ninguno de los estudios informó el uso de una lista de verificación estandarizada como lo describió Vitale 1996 o el American College of Sports Medicine (ACSM 2014). El uso de listas de verificación antes de comenzar la evaluación del estado físico y el entrenamiento proporciona una selección rigurosa para asegurar que se minimiza el riesgo de eventos adversos.

Se examinaron las tasas de abandonos y de adherencia para determinar cuán aceptable y viable es llevar a cabo un programa de acondicionamiento cardiorrespiratorio para los pacientes después del TCE. Los ocho estudios informaron una tasa de abandonos baja, lo que demuestra que esta intervención se consideró aceptable para los participantes. La adherencia se definió como el número de sesiones de ejercicios completadas en comparación con el número prescrito de sesiones de ejercicios. Cinco de los ocho estudios informaron las tasas de adherencia (Bateman 2001; Driver 2004; Driver 2006; Driver 2009; Laskin 2001), las mismos variaron de 69% en Bateman 2001 a 100% en Driver 2004 y Laskin 2001. En estos cinco estudios se supervisó cada sesión de ejercicios, por lo que las tasas de adherencia informadas deben ser exactas. El estudio con la adherencia más baja fue un estudio de pacientes hospitalizados y la adherencia deficiente se debió principalmente a que los participantes fueron dados de alta del hospital antes de finalizar el programa y tuvieron dificultades para regresar a cada sesión. Es interesante que, en los dos estudios con adherencia del 100% y ningún abandono, los participantes residían en la comunidad. El entrenamiento del estado físico prescrito después del alta, y el regreso a casa, parece por lo tanto aceptable y viable.

Se examinó la capacidad para lograr los parámetros de entrenamiento para determinar cuán apropiado es un programa de acondicionamiento cardiorrespiratorio para los pacientes después del TCE. Sin embargo, sólo un estudio informó los parámetros reales de entrenamiento logrados (Bateman 2001). En este estudio, los participantes del grupo de intervención pasaron en promedio 16 minutos por sesión dentro o por encima de la zona de la frecuencia cardíaca de entrenamiento y sólo realizaron ejercicios durante un promedio 24,5 minutos por sesión. Este estudio se realizó en los pacientes de forma temprana después del TCE (< seis meses) y hubo incertidumbre con respecto al efecto del entrenamiento físico en comparación con la terapia de relajación en el estado cardiorrespiratorio (DM 21,5 W, IC del 95%: ‐12,3 a 55,3). En comparación, en el programa de acondicionamiento físico para el estudio que reveló un efecto positivo claro, los participantes que habían sufrido la lesión hacía más de un año debían trabajar durante 60 minutos dentro de su zona de frecuencia cardíaca de entrenamiento. Sin embargo, debido a que Bateman 2001 no informó la intensidad real del entrenamiento ni la duración lograda, sólo es posible especular que una duración más larga del ejercicio a una intensidad suficiente puede haber contribuido con la diferencia en el tamaño del efecto entre estos dos estudios. En un estudio observacional realizado por el equipo fue posible demostrar que poco después del TCE (el tiempo promedio desde la lesión fue de 3,7 meses) los pacientes pueden lograr la dosificación del ejercicio recomendada en las guías del American College of Sports Medicine (ACSM) para los pacientes después de un accidente cerebrovascular o un TCE (Palmer‐McLean 2009). Sin embargo, lo anterior sólo pudo ser logrado mediante la participación en ejercicios de una duración más larga y de menor intensidad (52 minutos) debido a la dificultad para mantener volúmenes de trabajo de intensidad moderada (Hassett 2012). Este hecho indicaría la importancia en las situaciones clínicas de vigilar la intensidad y la duración del ejercicio para asegurar que se proporciona una dosificación suficiente del ejercicio, y destaca la necesidad de estudios futuros para vigilar y registrar los datos reales del entrenamiento sistemáticamente para cada uno de los parámetros y permitir la interpretación válida de los resultados.

Compleción y aplicabilidad general de las pruebas

Ninguno de los estudios incluidos investigó el efecto del entrenamiento físico en los niños que habían padecido un TCE. Esta es una brecha principal en la bibliografía cuando se considera que una de las metas de los equipos de rehabilitación de los niños es la reintegración exitosa de estos niños en el deporte y las actividades físicas (Rossi 1996). No fue posible realizar análisis de subgrupos debido a que hubo muy pocos estudios incluidos, por lo tanto, la generalización a la población con TCE es difícil. La mayoría de los participantes en los ocho estudios incluidos era hombres adultos ambulatorios que habían tenido un TCE grave, por lo que los resultados no se pueden aplicar a otros subgrupos de TCE (p. ej. mujeres adolescentes con un TCE leve). Uno de los análisis de subgrupos planificados fue el momento desde la lesión. Los resultados diferentes entre los estudios Bateman 2001 y Driver 2004 para el estado cardiorrespiratorio, así como el análisis post hoc que incluyó sólo a los participantes con TCE crónico para la depresión, sugieren que el tiempo luego de la lesión puede ser importante. Los estudios futuros que investigan los efectos del entrenamiento físico en el estadio agudo y crónico después del TCE permitirían la realización de análisis de subgrupos para determinar si el entrenamiento físico es en verdad efectivo para mejorar el estado cardiorrespiratorio, la depresión y otros resultados de forma temprana y tardía después del TCE.

Calidad de la evidencia

Esta revisión incluyó sólo ensayos controlados aleatorios, por lo tanto, cada resultado se consideró inicialmente como evidencia de alta calidad. Al aplicar los criterios GRADE, un resultado se disminuyó a evidencia de calidad moderada (función neuromotora), cuatro a evidencia de baja calidad (estado cardiorrespiratorio, composición corporal, análisis de sensibilidad de la depresión, abandonos) y tres se disminuyeron a evidencia de muy baja calidad (fatiga, depresión, fuerza). La evidencia de disminuyó debido al riesgo de sesgo del diseño de estudio, la heterogeneidad entre los estudios y la imprecisión del cálculo del efecto. Para el resultado primario (estado cardiorrespiratorio), los dos estudios que proporcionaron la mayoría de los datos tuvieron diferentes tamaños del efecto. Se ha formulado la hipótesis de que una de las razones de esta diferencia fue la diferencia en el tiempo luego de la lesión de los participantes. Sin embargo, el estudio con el tamaño más grande del efecto no usó la ocultación de la asignación, lo cual se sabe que aumenta el tamaño del efecto de los ensayos en comparación con los ensayos con una ocultación adecuada de la asignación (Savovic 2012). Este hecho podría haber contribuido a la diferencia en los tamaños del efecto. Lo anterior refuerza la necesidad de que los ensayos futuros incorporen características de diseño que se sabe que disminuyen el sesgo.

Sesgos potenciales en el proceso de revisión

La estrategia de búsqueda fue minuciosa e incorporó 10 bases de datos electrónicas, la International Clinical Trials Registry Platform (la base de datos central para 16 registros de ensayos), una revisión de las listas de referencias de las revisiones sistemáticas relevantes y cualquier artículo de copia completa obtenido, y la comunicación personal con los autores principales de los estudios incluidos. Se excluyeron cinco estudios debido a que no fue posible confirmar su elegibilidad (Froehlich‐Grobe 2004; Herterich 2002; Kurowski 2017; Rarick 1968; Zhu 2004) y aún se debe confirmar la elegibilidad de un estudio (McClanachan 2013), por lo tanto, no es posible tener la seguridad de que no se han omitido algunos datos. Sin embargo, para cuatro de los estudios que fueron excluidos, no se conoce si se incluyó algún participante con TCE, y para cuatro de los cinco estudios no se conoce si la intervención cumplió con los criterios para el entrenamiento físico. Como tal, se cree que es probable que estos estudios hubieran sido excluidos o hubieran contribuido con pocos participantes a la revisión. Aunque no se realizaron búsquedas manuales, el Grupo Cochrane de Lesiones emprende la búsqueda manual; por lo tanto cualquier, cualquier referencia relevante adicional debería haberse identificado a partir del Registro de Ensayos del Grupo Cochrane de Lesiones. Por lo tanto, se tiene la seguridad de que esta revisión contiene todos los estudios pertinentes sobre este tema.

Acuerdos y desacuerdos con otros estudios o revisiones

No ha habido ninguna otra revisión sistemática publicada que haya evaluado el entrenamiento físico en pacientes con TCE, aunque hay revisiones sistemáticas que han evaluado el entrenamiento físico en una población similar (es decir pacientes después del accidente cerebrovascular; Saunders 2016) y en poblaciones con deficiencias similares (p.ej. pacientes con depresión; Cooney 2013). Los resultados de la revisión para los pacientes después del accidente cerebrovascular fueron similares a los de esta revisión; hubo una leve mejoría del estado cardiorrespiratorio con el entrenamiento al final de la intervención (captación máxima de oxígeno [VO2 máximo] DM 2,86; IC del 95%: 1,76 a 3,96 ml/kg/min; 9 estudios, 425 participantes) y resultados contradictorios para otros resultados, como mejorías pequeñas en la velocidad de caminata y la resistencia y efectos poco claros sobre la depresión (Saunders 2016). Del mismo modo que con esta revisión, el número pequeño de estudios, la cantidad escasa de participantes y la variedad de medidas de resultado utilizadas limitó las conclusiones que podrían establecerse para muchos de los resultados secundarios. Una revisión Cochrane que evaluó el efecto del ejercicio sobre la depresión tanto en poblaciones clínicas como no clínicas ‐ 33 estudios usaron entrenamiento físico ‐ informó un efecto moderado del ejercicio en la depresión (DME ‐0,62; IC del 95%: ‐0,81 a ‐0,42; 35 estudios, 1356 participantes) (Cooney 2013). Sin embargo, un análisis de sensibilidad que incorporó sólo los estudios en bajo riesgo de sesgo mostró una dilución de este efecto, de manera que todavía hay incertidumbre con respecto al tamaño del efecto del entrenamiento físico en la depresión.

Figure 1

Study flow diagram

Figure 2

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

Figure 3

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

Analysis 1.1

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 1 Cardiorespiratory fitness ‐ peak power output.

Analysis 1.2

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 2 Body composition ‐ standardised mean difference.

Analysis 1.3

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 3 Strength ‐ standardised mean difference.

Analysis 1.4

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 4 Fatigue‐standardised mean difference.

Analysis 1.5

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 5 Depression ‐ standardised mean difference.

Analysis 1.6

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 6 Neuromotor function ‐ metre/s.

Analysis 1.7

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 7 Quality of life ‐ standardised mean difference.

Analysis 1.8

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 8 Dropouts during the intervention phase.

Analysis 1.9

Comparison 1 Exercise programme prescribed alone versus a non‐exercise intervention or no intervention, Outcome 9 Depression post hoc analysis (studies with sub‐acute to chronic participants at home) ‐ standardised mean difference.

Cardiorespiratory exercise programmes prescribed alone compared with usual care, a non‐exercise intervention, or no intervention for people after traumatic brain injury
Patient or population: people with traumatic brain injury of any age or severity Setting: any setting, including inpatient, outpatient, community and home Intervention: exercise using large muscle groups prescribed at least three times a week for a duration of at least 20 minutes at an intensity that represented a clear attempt to improve cardiorespiratory fitness Comparison: usual care, a non‐exercise intervention, or no intervention
Outcomes	*Illustrative comparative risks (95% CI)**		No of participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	No fitness training	Fitness training
Cardiorespiratory fitness ‐ peak power output (W) ‐ end of intervention	The mean peak power output ranged across control groups from 85 W to 145 W	The mean peak power output in the intervention groups was on average 35 W higher (95% CI 3 to 68)	67 (3 studies)	⊕⊕ low^a,b	The MCID has not been determined for this outcome in this or a similar population, however in people with severe COPD, the MCID is 4 W (Puhan 2011).
Depression ‐ SMD ‐ end of intervention	The mean depression score ranged across control groups from 1.1 to 21.2 points on various depression scales	The SMD between groups was on average 0.43 lower in the intervention group (95% CI 0.92 to ‐0.06)	220 (4 studies)	⊕ very low ^a,b,c	An SMD of 0.43 represents a difference between groups in scores on the depression subscale of the Hospital Anxiety Depression Scale (0 to 21) of a 1.7 point reduction in depression with the 95% confidence intervals indicating an effect between a 3.7 reduction in depression to a 0.2 point increase in depression. Post hoc analysis pooling the data from the 3 studies that included subacute and chronic participants at home, the SMD indicated a moderate reduction in depression with fitness training with the 95% CIs indicating between a large and small reduction in depression with fitness training compared to the control interventions (SMD ‐0.61, 95% CI ‐1.1 to ‐0.11; 3 studies, 178 participants; low‐quality evidence).
Cognition	See comment	See comment	111 (2 studies)	See comment	Data could not be pooled because the outcome measures from both studies did not measure the same domains of cognition. There were no immediate or long‐term effects in either study.
Fatigue ‐ SMD ‐ end of intervention	The mean fatigue score ranged among control groups from 1.3 to 6.6 points on various fatigue scales	The SMD between groups was on average 0.32 lower in the intervention group (95% CI 0.90 to ‐0.26)	130 (3 studies)	⊕ very low ^a,b,c	An SMD of 0.32 represents a difference between groups in scores on a VAS fatigue scale (0 to 10) of approximately 0.9 points (95% CI ‐0.7 to 2.6), with a lower score indicating a reduction in fatigue.
Community integration	See comment	See comment	76 (1 study)	See comment	Data only available from one study; no clear evidence of effect.
Safety	See comment	See comment	183 (5 studies)	See comment	Trialists reported no adverse events occurred during the intervention phase in these studies.
The corresponding risk* (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; COPD: chronic obstructive pulmonary disease; MCID: minimal clinically important difference; SMD: standardised mean difference; VAS: visual analogue scale; W: watts
GRADE Working Group grades of evidence High quality: further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: we are very uncertain about the estimate.
^a Downgraded once because of risk of bias. ^b Downgraded once because of imprecision of results. ^c Downgraded once because of heterogeneity among included studies.

Table 1. Summary of the prescribed exercise programmes

Study	Setting	Supervised	Exercise type	Resistance exercises	Intensity	Duration	Frequency	Length of programme
Bateman 2001	Inpatient	Yes	Cycle ergometer	No	60% to 80% predicted heart rate maximum (210 − 0.65 x age)	Aim 30 minutes; average time 24.5 minutes	Aim 3 x week, average 2.1 x week	12 weeks
Driver 2004 Driver 2006 Driver 2009	Community	Yes	Aquatic	Yes	50% to 70% predicted heart rate maximum (Karvonen method)	60 minutes	3 x week	8 weeks
Hoffman 2010	1 x week community; ≥ 4 x week at home	Yes for 10 sessions, independent for 40 sessions	Participant choice (e.g. treadmill, cycle ergometer, indoor track, rowing machine, stair‐stepper)	No	60% heart rate max (220 − age) or 12/20 RPE	15 minutes education, 30 minutes warm‐up and cool‐down, 30 minutes aerobic	1 x supervised; ≥ 4 x independent	10 weeks
Laskin 2001	Outpatient	Yes	Cycle (or arm crank) ergometer	Yes	Heart rate and power output set to second ventilatory threshold from peak test	Average time 15 minutes week 1 to 28.5 minutes week 12 for continuous fitness component	3 x week	12 weeks
McMillan 2002	Home	Yes for 5 sessions, independent for 15 sessions	Basic fitness exercises, no equipment	Unknown	Not specified	45 minutes	5 x week	4 weeks
Salazar 2000	Inpatient	Yes	Fitness group	Unknown	At own pace	60 minutes	5 x week	8 weeks
Abbreviation RPE = rated perceived exertion

Table 1. Summary of the prescribed exercise programmes

Table 2. Justification of GRADE ratings for pooled outcomes

Outcome	Reasons to downgrade the evidence					Reasons to upgrade the evidence	GRADE rating
Outcome	Risk of bias of study design	Indirectness of evidence	Heterogeneity	Imprecision	Publication bias	Large magnitude of effect, dose response, effect of all plausible confounding factors	GRADE rating
Cardiorespiratory fitness (3 studies, 67 participants)	Downgraded x 1 level: allocation concealment at high risk of bias for 1 study; no blinded assessor for 1 study	Nil: not gold standard measure, but still common measure of fitness	Nil: I² = 36%; Chi² P = 0.21	Downgraded x 1 level: wide CI and 2 studies with positive and 1 study with negative mean differences	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 2 levels to LOW
Body composition (2 studies, 61 participants)	Downgraded x 1 level: allocation concealment high for 1 of 2 studies	Nil	Nil: I² = 0% Chi² P = 0.88	Downgraded x 1 level: wide CI	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 2 levels to LOW
Strength (2 studies, 23 participants)	Downgraded x 2 levels: allocation concealment high for 1 of 2 studies; unblinded assessor for 1 of 2 studies; small number of participants	Nil	Nil: I² = 0% Chi² P = 0.77	Downgraded x 1 level: wide CI and 1 study with positive and 1 study with negative mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 3 levels to VERY LOW
Fatigue (3 studies, 130 participants)	Downgraded x 1 level: allocation concealment unclear for 1 study and low for 2 studies; high risk (1 study) or unclear risk (2 studies) of detection bias: participants not blinded for self‐reported measure	Nil	Downgraded x 1 level: I² = 56% Chi² P = 0.11	Downgraded x 1: wide CI and 2 studies with negative and 1 study with positive mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 3 levels to VERY LOW
Depression (4 studies, 220 participants)	Downgraded x 1 level: allocation concealment high for 1 of 4 studies and unclear for 1 of 4 studies; high risk (1 study) or unclear risk (3 studies) of detection bias: participants not blinded for self‐reported measure	Nil	Downgraded x 1 level: I² = 64% Chi² P = 0.04	Downgraded x 1: wide CI and 3 studies with negative and 1 study with positive mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 3 levels to VERY LOW
Depression post hoc analysis (3 studies, 178 participants)	Downgraded x 1 level: allocation concealment high for 1 of 4 studies and unclear for 1 of 4 studies; high risk (1 study) or unclear risk (3 studies) of detection bias: participants not blinded for self‐reported measure	Nil	Downgraded x 1 level: I² = 53% Chi² P = 0.04	Nil	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 2 levels to LOW
Neuromotor function (2 studies, 109 participants)	Nil	Nil	Nil: I² = 28% Chi² P = 0.24	Downgraded x 1 level: wide CI and 1 study with positive and 1 study with negative mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 1 level to MODERATE
Acceptance ‐ dropout rate (3 studies, 222 participants)	Downgraded x 1 level: allocation concealment high for 1 of 3 studies	Nil	Downgraded x 1 level: I² = 50% Chi² P = 0.13	Downgraded x 1 level: wide CI and 2 studies with negative and 1 study with positive mean difference	Nil: no published protocol, but no bias serious enough to downgrade	Nil	Downgraded 3 levels to VERY LOW
Abbreviation CI = confidence interval

Table 2. Justification of GRADE ratings for pooled outcomes

Table 3. Additional data for comparison 1

Outcome	Study	Outcome measure	Timing	Intervention group Mean (SD); N	Control group Mean (SD); N	Mean between group difference (MD), 95% CI
Cognition	McMillan 2002	PASAT	End of intervention	37 (13); 38	32 (14); 48	MD 5, 95% CI ‐0.7 to 10.7
		PASAT	End of follow‐up	38 (14); 35	34 (15); 38	MD 4, 95% CI ‐2.7 to 10.7
		Sunderland Memory Questionnaire	End of intervention	52 (35); 38	52 (37); 48	MD 0, 95% CI ‐15.3 to 15.3
		Sunderland Memory Questionnaire	End of follow‐up	49 (36); 35	47 (34); 38	MD 2, 95% CI ‐14.1 to 18.1
		Adult Memory & Information Processing Battery	End of intervention	24 (12); 38	23 (12); 48	MD 1, 95% CI ‐4.1 to 6.1
		Adult Memory & Information Processing Battery	End of follow‐up	34 (10); 35	31 (13); 38	MD 3, 95% CI ‐2.3 to 8.3
	Driver 2009	POMS: confusion subscale	End of intervention	0.6 (0.3); 8	0.9 (0.6); 8	MD ‐0.2, 95% CI‐0.7 to 0.2
Community integration	Hoffman 2010	CHART‐SF	End of intervention	85 (14); 37	83 (12); 39	MD 1.7, 95% CI ‐4.1 to 7.5
Global measure of function	Bateman 2001	Barthel Index	End of intervention	17 (4); 24	18 (3); 22	MD ‐0.9, 95% CI ‐2.9 to 1.2
Global measure of function	Bateman 2001	Barthel Index	End of follow‐up	18 (4); 22	19 (3); 21	MD ‐1.1, 95% CI ‐3.3 to 1.2
Abbreviations CHART‐SF = Craig Handicap Assessment and Reporting Technique ‐ Short Form; PASAT = Paced Auditory Serial Addition Test; POMS = Profile of Mood States

Table 3. Additional data for comparison 1

Table 4. Additional data for comparison 2

Outcome	Study	Outcome measure	Timing	Intervention group Mean (SD); N	Control group Mean (SD); N	Between‐group difference
Cognition	Salazar 2000	Buschke Selective Reminding Test	End of follow‐up	67 (34); 42	63 (40); 34	MD 4.0, 95% CI ‐12.9 to 20.9
Cognition		PASAT	End of follow‐up	42 (14); 42	38 (12); 33	MD 3.5, 95% CI ‐2.3 to 9.4
Depression		Diagnosis of major depression	End of follow‐up	7/42^a	9/34^a	OR 0.6, 95% CI 0.2 to 1.7
Community integration		Return to work	End of follow‐up	60/67^a	50/53^a	OR 0.5, 95% CI 0.1 to 2.1
Acceptance		Dropouts	During the intervention phase	7/67^a	2/53^a	OR 3.0, 95% CI 0.6 to 15.0
^aNumber of events over total number of cases Abbreviations MD = mean difference; OR = odds ratio; PASAT = Paced Auditory Serial Addition Test

Table 4. Additional data for comparison 2

Comparison 1. Exercise programme prescribed alone versus a non‐exercise intervention or no intervention

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Cardiorespiratory fitness ‐ peak power output Show forest plot	3		Mean Difference (IV, Random, 95% CI)	Subtotals only

1.1 End of intervention	3	67	Mean Difference (IV, Random, 95% CI)	35.47 [2.53, 68.41]
1.2 End of follow‐up	1	40	Mean Difference (IV, Random, 95% CI)	3.27 [‐34.21, 40.75]
2 Body composition ‐ standardised mean difference Show forest plot	2		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 End of intervention	2	61	Std. Mean Difference (IV, Random, 95% CI)	0.29 [‐0.22, 0.79]
2.2 End of follow‐up	1	41	Std. Mean Difference (IV, Random, 95% CI)	0.52 [‐0.11, 1.14]
3 Strength ‐ standardised mean difference Show forest plot	2	23	Std. Mean Difference (IV, Random, 95% CI)	‐0.02 [‐0.86, 0.83]

4 Fatigue‐standardised mean difference Show forest plot	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

4.1 End of intervention	3	130	Std. Mean Difference (IV, Random, 95% CI)	‐0.32 [‐0.90, 0.26]
4.2 End of follow‐up	1	40	Std. Mean Difference (IV, Random, 95% CI)	0.35 [‐0.28, 0.97]
5 Depression ‐ standardised mean difference Show forest plot	4		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

5.1 End of intervention	4	220	Std. Mean Difference (IV, Random, 95% CI)	‐0.43 [‐0.92, 0.06]
5.2 End of follow‐up	2	113	Std. Mean Difference (IV, Random, 95% CI)	‐0.07 [‐0.84, 0.69]
6 Neuromotor function ‐ metre/s Show forest plot	2	109	Mean Difference (IV, Random, 95% CI)	0.01 [‐0.25, 0.27]

7 Quality of life ‐ standardised mean difference Show forest plot	3		Std. Mean Difference (IV, Random, 95% CI)	Totals not selected

7.1 End of intervention	3		Std. Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
7.2 End of follow‐up	1		Std. Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
8 Dropouts during the intervention phase Show forest plot	3	222	Odds Ratio (M‐H, Random, 95% CI)	3.14 [0.32, 30.80]

9 Depression post hoc analysis (studies with sub‐acute to chronic participants at home) ‐ standardised mean difference Show forest plot	3		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only

9.1 End of intervention	3	178	Std. Mean Difference (IV, Random, 95% CI)	‐0.61 [‐1.10, ‐0.11]

Comparison 1. Exercise programme prescribed alone versus a non‐exercise intervention or no intervention