Currently, complete recovery is unattainable for most individuals with spinal cord injury (SCI). Instead, recovery is typically accompanied by persistent sensory and motor deficits. Restoration of preinjury function will likely depend on improving plasticity and integration of these impaired systems. Eccentric muscle actions require precise integration of sensorimotor signals and are predominant during the yield (E2) phase of locomotion. Motor neuron activation and control during eccentric contractions is impaired across a number of central nervous system (CNS) disorders, but remains unexamined after SCI. Therefore, we characterized locomotor recovery after contusive SCI using hindlimb (HL) kinematics and electromyographic (EMG) recordings with specific consideration of eccentric phases of treadmill (TM) walking. Deficits in E2 and a caudal shift of locomotor subphases persisted throughout the 3-week recovery period. EMG records showed notable deficits in the semitendinosus (ST) during yield. Unlike other HL muscles, recruitment of ST changed with recovery. At 7 days, the typical dual-burst pattern of ST was lost and the second burst (ST2) was indistinct. By 21 days, the dual-burst pattern returned, but latencies remained impaired. We show that ST2 burst duration is highly predictive of open field Basso, Beattie, Bresnahan (BBB) scores. Moreover, we found that simple changes in locomotor specificity which enhance eccentric actions result in new motor patterns after SCI. Our findings identify a caudal shift in stepping kinematics, irregularities in E2, and aberrant ST2 bursting as markers of incomplete recovery. These residual impairments may provide opportunities for targeted rehabilitation.
Keywords: Kinematics, locomotion, rehabilitation, spinal cord injury
Spinal cord injury (SCI) results in a diverse range of behavioral outcomes that depend on the type, severity, and level of injury. To date, the extent of recovered central nervous system (CNS) control over locomotion has been best elucidated in reductionistic lesion models (Kaegi et al. 2002; Ballermann et al. 2006; Johnson et al. 2012). Surprisingly, less is understood about recovery from contusion-type lesions, which replicate human SCI. Contusive SCI results in complex pathology with distinct anatomical, behavioral, and cellular sequella along the neuraxis (Stokes and Jakeman 2002; Profyris et al. 2004; Detloff et al. 2008). It is well-accepted that greater sparing of descending midbrain/brainstem pathways improve motor function after contusion (Fehlings and Tator 1995; Basso et al. 2002; Schucht et al. 2002). However, factors that promote supraspinal and afferent integration during locomotion have received little attention.
Differential recovery after contusive SCI may be identified by changes in gait biomechanics and muscle activation patterns. After hemisection, postural elevation, interlimb uncoupling, and aberrant coactivation patterns between adjacent muscles persist and indicate the limits of recovery (Kaegi et al. 2002; Ballermann et al. 2006). Given the compensatory nature of this injury, it is unclear whether similar factors delineate recovery after bilateral contusion. We previously identified at least one motor feature that remains impaired after SCI – the yield phase during weight acceptance (Basso et al. 1994). Here, we ask whether the kinematics or electromyographic (EMG) metrics of yield may be associated with the extent of recovery.
Eccentric motor control represents a hallmark of skilled locomotion that is impaired across CNS injury models, but remains unexamined after contusive SCI (Helgren and Goldberger 1993; Basso et al. 1994; Damiano et al. 2001; Dibble et al. 2009). In this ubiquitous action, motor units are partially recruited to keep muscle force below the external load. To attain effective eccentric muscle lengthening, descending drive is precisely controlled to match the afferent input of the movement (Enoka 1996). A predominant eccentric period in the step cycle occurs prior to ground contact and during weight acceptance, when hamstring muscles like the semitendinosus (ST) lengthen to decelerate the hindlimb (HL) and dissipate impact forces during yield (E2). Importantly, recruitment of ST adapts to a variety of locomotor conditions and requires descending control for optimal function (Buford et al. 1990; Pratt et al. 1996; Smith et al. 1998). Our previous work in the cat shows that the eccentric phase of locomotion remains impaired despite marked recovery from a hemisection (Basso et al. 1994). To further this observation and identify mechanisms of eccentric control after contusion, we examined ST recruitment patterns over time and at recovery plateau. Whether poor eccentric activity in ST or other HL muscles prevents optimal recovery is unknown.
The present study was designed to identify features of recovered walking patterns that differentiate functional restitution after a mild/moderate, midthoracic contusion injury. Detailed assessment of HL muscle recruitment and joint kinematics describe the extent of motor control. Our findings suggest that eccentric actions of ST provide novel insight into mechanisms of locomotor recovery after SCI.
Materials and Methods
Subjects and surgeries
Experiments were conducted in 14 female Sprague-Dawley rats (250–300 g, Harlan, Indianapolis, Indiana) that were randomly assigned to control laminectomy (LAM) or SCI groups following EMG implantation. Naive data collection for all rats served as baseline. Comparisons included Naive (n = 14), LAM (n = 5), and SCI (n = 9). Animals were housed 2–3 per cage in a controlled environment (12 h light/dark cycle) with food and water available ad libitum. Housing, surgical procedures, and assessment of behavior was done in accordance with The Ohio State University Laboratory Animal Care and Use Committee. For all surgeries, rats were anesthetized intraperitoneal (i.p.) with ketamine (80 mg/kg) and xylazine (20 mg/kg). During each surgical procedure, a heating pad maintained body temperature. Prophylactic antibiotics (gentomycin sulfate 1 mg/kg) and saline were given post surgery to prevent infection and dehydration.
Subjects were acclimated to the treadmill (TM) and trained to walk steadily prior to EMG implantation; this training required 2–3 weeks. During the first surgery, bipolar EMG electrodes were implanted into the tibialis anterior (TA), lateral gastrocnemius (LG), and the ST of the left HL. These muscles were selected based on electrode stability and locomotor biomechanics. A small rostral–caudal incision was made along the sagittal suture of the skull and four screws were anchored on each side. Teflon-coated, multistranded stainless steel wires fixed to a head plug were routed subcutaneously to the HL and implanted into exposed muscles with a hypodermic needle. Electrode functionality was confirmed by electrical stimulation through each lead (∼0.2–0.8 mA, 0.2 msec cathodal pulse) to elicit a muscle twitch. A ground electrode remained subcutaneous to serve as reference. The head connector was cemented with varnish and dental acrylic to the screws, and incisions were closed with suture.
Spinal cord injury
In the time between EMG implant and SCI (11 days), normal open field locomotion was confirmed for each rat. Over this period, rats were reacclimated to the TM and learned to walk with the EMG wire connected to the head plug. This data collection was used for naive comparison. In the second surgery, rats were anesthetized as described previously, and a midthoracic T8 laminectomy exposed the spinal cord. Animals randomized to the SCI group received a mild/moderate injury produced by rapidly impacting the spinal cord using the OSU Electromagnetic Spinal Cord Injury Device or the Infinite Horizons (IH) Device (Stokes et al. 1992; Stokes and Jakeman 2002). Following contusion or LAM control, dorsal musculature was sutured and skin was closed using surgical clips. Sterile saline was administered subcutaneously to prevent dehydration. Antibiotics were delivered daily and bladders were manually expressed 2×/day until the bladder reflex returned. Vitamin C pellets were given to prevent urinary tract infections (Behrmann et al. 1992). Animals that exhibited wiring problems or bladder infection following surgery were not used for EMG collection (n = 2).
To examine muscle recruitment patterns after SCI, EMG signals were recorded and synchronized with joint kinematics for six animals and averaged across at least 20 steps on the TM (Columbus Instruments, Columbus, Ohio). For downhill recordings, the TM belt was set to a 10% (5.7 degrees) downslope grade. Flexible insulated cables were attached to a head plug, and connected via a commutator to the amplifier, allowing free movement of the subjects on the TM belt. A sugar water dispenser at the front of the belt prompted forward locomotion. Preoperative training frequency and duration was adjusted per rat until long bouts of sustained stepping occurred while drinking. Postoperatively, brief exposure to the TM occurred to maintain comfort with the task. Collection occurred at the same speed (12 m/min) and while drinking to eliminate backward drift.
The EMG signals were amplified at a gain of 1K with an AM-Systems model 1700 differential amplifier. The bandpass filters were set for 20 Hz–5 KHz, and a 60-Hz notch filter was engaged. Computerized data acquisition was accomplished with a sampling rate of 2 KHz using either a 16-bit Datapac 2K2 system (Run Technologies, Laguna Hills, California) or a 16-bit CED Power 1401 system with Spike2 software (CED, Cambridge, UK). The EMG records were adjusted to remove DC offsets, rectified, and averaged across 20+ steps off-line using a custom script that used initial contact times as a triggering event. A burst detection program determined the beginning (onset) and end (offset) of each EMG burst and calculated relative to initial contact by determining when the EMG level crossed a threshold set to 2 standard deviations above the mean activity level during quiescence for each muscle. Visual inspection was used to adjust onset and offset times as required to eliminate spurious bursts and locate the main burst periods associated with locomotion. Burst durations were calculated based on the onset and offset times. Digital video records were synchronized with the EMG recordings by means of an LED light that was visible to the camera, with the voltage pulse for the light recorded along with the EMG.
Locomotor recovery was assessed using the 21-point Basso, Beattie, Bresnahan (BBB) locomotor rating scale (Basso et al. 1995). Scores range from no HL movement (0) to normal locomotor function (21). Rating criteria considered joint movement, weight support, plantar stepping, coordination, toe clearance, paw position, as well as trunk and tail control. Open field activity of each rat occurred for 4 min by two raters blind to group assignment. Assessments were done prior to injury, at 1 and 7 days postoperatively (dpo), and weekly thereafter.
All rats had two-dimensional (2D) kinematic recordings of TM walking before and 3 weeks after SCI. Left HLs were shaved and bony prominences were marked with permanent marker preoperatively. The prominences included the iliac crest, greater trochanter, femoral condyle, lateral malleolus, and head of the fifth metatarsal. A videotape record of quadrupedal locomotion (10–20 step bouts) was collected using a Panasonic WV-CL350 camera (60 Hz) with a time-code generator. The same LED light used to synchronize the EMG and digital video records was visible to the analogue video camera and was used to synchronize the records. HL kinematic markers were digitized using PEAK Motus. To account for movement of the knee joint, a triangulation program was used to estimate its position (Goslow et al. 1973). Actual femur and tibia bone lengths were collected at sacrifice and used with the hip and ankle X, Y positions to derive location. Angular excursions were calculated for the hip, knee, and ankle during each phase of quadruped gait: Initial Contact (E1), Yield (E2), Lift Off (E3), and Peak Flexion (F) (Basso et al. 1994). Timing of initial contact along with the LED synchronization light served as the reference times to synchronize EMG and kinematic data. Angle–angle diagrams were constructed by plotting joint excursions (hip–knee or knee–ankle) against one another to assess intralimb coordination.
Rats were perfused with 0.1 M phosphate buffered saline followed by 4% paraformaldehyde. Tissue was collected and cryoprotected in sucrose. The lesion site was transversely sectioned (20 μm) and stained for myelin using eriochrome cyanine. The section with the largest lesion and least amount of stained white matter represented the lesion epicenter. Area of stained white matter at the epicenter was divided by the total cross-sectional area of an uninjured cord at the same vertebral level to serve as a measure of injury severity (Kloos et al. 2005).
All outcome measures were analyzed compared to naive. Kinematic comparisons were done using a repeated measures analysis of variance (ANOVA) and Tukey's post hoc test. Significance observed with BBB scores was determined using a Mann–Whitney U-test to account for unequal sample size. Correlations between EMG burst duration, BBB score, and white matter sparing were done using Pearson's correlation analysis. Significance was set at P < 0.05 and mean ± SEM are shown.
Residual deficits contribute to a new walking strategy after mild SCI
Using the BBB scale, spontaneous recovery occurred over 21 days after mild SCI but residual impairments prevented normal locomotion (Fig. 1). Mild SCI resulted in severe paresis with slight and extensive HL movements 1 day after SCI (Mean BBB = 6.83 ± 0.655). Weight supported stepping recovered within 7 days. Despite rapid improvement, recovery plateaued at levels significantly below normal at 21 days (Mean BBB = 15.75 ± 1.085; P < 0.05). While one animal attained near normal locomotion (BBB = 19), remaining animals had persistent trunk instability (100%), toe dragging (37.5%), and paw rotation at lift off (100%) or initial contact (37.5%).
Open field locomotion. Spontaneous recovery occurred in the open field after mild SCI. BBB scores plateaued by 21 days and remained significantly lower than control (mean SCI = 15.7 ± 1.085). Residual deficits at 21 days included toe dragging,...
Using 2D TM kinematics, we quantified the plateaued walking behavior across subphases of locomotion (Fig. 2; Basso et al. 1994). Hip movements are biphasic and include two subphases, flexion (F) and extension (E). Knee and ankle movements are more complex and are divided into four subphases (E1, E2, E3, F). The first extension subphase (E1) occurs from peak flexion in swing until initial paw contact on the ground. The E2 subphase, from initial contact through weight acceptance, represents joint flexion during yield and relies on eccentric muscle lengthening. During E3, midstance to lift off, all joints extend. Lift off to peak flexion represents the flexion (F) subphase. Thus, stance includes E2 and E3 and swing includes F and E1 (Fig. 2).
Stick figure diagrams at the end of each phase of gait illustrate prolonged extension during TM locomotion. Subphases of locomotion include E1, E2, E3, and F. The third extension phase (E3) occurs from midstance to lift off where knee and ankle extension...
After recovery from SCI, the position of the paw relative to the pelvis showed significant caudal displacement during all phases of gait (Fig. 2). The caudal shift for injured rats (dotted lines) was 2.39 ± 0.23 cm (P < 0.01) at lift off and 1.24 ± 0.29 cm (P < 0.01) at initial contact compared to naive (solid lines; Fig. 2). During E1, a 43% reduction in forward swing occurred after SCI (3.35 ± 0.473 cm, SCI; 5.88 ± 0.488 cm, Naive; P < 0.05). This caudal shift was reflected in significant differences in angular excursion of all HL joints (Fig. 3). Knee and ankle extension decreased during late swing (E1) and yield (E2) (P < 0.05). Significantly greater extension occurred in the hip, knee, and ankle during late stance (E3), leading to more excursion during flexion (F) after SCI (P < 0.05). The increase in flexion was not due to hypermetria as toe height was reduced after injury (toe height: 1.88 ± 0.151 cm, SCI; 2.10 ± 0.174 cm, Naive; Fig. 2); rather, greater flexion represented the return from prolonged extension at lift off. At lift off, the pelvis was on average 0.78 cm higher after SCI in 60% of animals. Implantation of EMG electrodes did not affect joint angular excursion (compare Naive and LAM groups, Fig. 3).
Angular excursion profiles of hip, knee, and ankle joints. Precise kinematic analysis of joint excursion between different phases of gait reveal altered biomechanics after SCI. Extension of the knee and ankle significantly increased from late stance to...
Recovery of intralimb coordination occurs in a proximal to distal manner
To examine coordinated movement between HL joints during locomotion, angle–angle diagrams were constructed by plotting the excursion of one joint against another. Coordination between proximal (hip–knee) or distal (knee–ankle) joints was compared to determine the extent of recovery. Angle–angle diagrams display joint excursion, position of the joints during excursion, and the coordination between joints (Basso et al. 1994). In normal locomotion, a curvilinear shape emerges when one joint moves to a greater extent (more excursion) than the other joint (Fig. 4). Fine motor control is made evident by fractionated movement, or independent control of joints. Fractionation is most clearly demonstrated in E2, where HL joints are required to flex while another extends. Intralimb coordination results when a reproducible and precise curvilinear pattern of movement occurs over multiple step cycles.
Fractionated movement in proximal and distal joints. Angle–angle plots were used to describe intralimb coordination between proximal (hip–knee) or distal (knee–ankle) joints. Naive plots depict curvilinear patterns between the...
After recovery from SCI, coordination between distal (knee and ankle) joints is most impaired. Linear rather than curvilinear paths depict poor fractionated joint movements. The linear pattern during stance results from a lack of E2 or yield phase (Fig. 4). The knee and ankle had equivalent changes in excursion and did not flex or extend in opposition to each other. Proximal coordination between the hip and knee was less impaired but a change in shape and position of the angle–angle plot was apparent (Fig. 4). A second flexion occurred at the knee during E2 (arrow, Fig. 4). A double yield was observed in 55% of animals. Prolonged extension is evident by the rightward and upward shift in position of the post op hip–knee and knee–ankle plots. At E3, the hip becomes approximately two times more extended than the knee, demonstrating greater proximal extension (Fig. 4).
Joint kinematics and timing of muscle activity
In naive animals, TA onset occurs with ankle dorsiflexion while LG onset occurs with plantar flexion before ground contact (Fig. 5). Both muscles are briefly coactive during terminal swing. TA offset occurs prior to plantar flexion and E1 (mean duration = 210.8 msec), and LG remains active during stance (mean duration = 442.9 msec). The dual-burst pattern of ST coincides with extension and flexion in the hip and knee. Onset of ST1 occurs during hip extension (mean duration = 156.8 msec) and ST2 during knee flexion through weight acceptance (mean duration = 248.2 msec). The double burst is separated by a brief pause during E1 while the hip flexes and the knee extends in midswing to move the paw forward.
Comparison of HL muscle activity with changes in angular kinematics before and after SCI. EMG activity is aligned with kinematics of the hip, knee, and ankle in the same animal before and 21 days after mild SCI. The vertical line marks stance onset. Black...
Timing and overall pattern of muscle recruitment changed after injury alongside altered joint kinematics. At the ankle, marked changes were evident compared to naive that were maintained throughout recovery. At 21 days, plantar flexion is absent at the ankle and LG onset instead occurs during a period of prolonged dorsiflexion before ground contact (Fig. 5). A reduction in burst duration is apparent in both muscles relative to naive-TA (–25.6 ± 7.5%); LG (–44.1 ± 12.0%). These reductions were independent of recovery in the open field (Fig. 6).
Activation patterns of the semitendinosis change with recovery. EMG activity is plotted in the same animal over time. The vertical line marks stance onset. Seven days after SCI, forelimb–hindlimb coordination and plantar stepping was not consistent...
Activity of ST changed over time but did not return to normal by 21 days. Early after SCI, with only frequent stepping and limited forelimb–hindlimb coordination (BBB = 12) at 7 days, the dual-burst pattern of the ST is lost and only a single prolonged burst occurs. Dual bursts return by 21 days when coordination and stepping frequency recover (BBB = 15; Fig. 6). ST1 fires later throughout recovery and occurs ∼101.9 msec closer to initial contact, and for shorter duration (–11.3 ± 24.5%) compared to naive (Fig. 7). After recovery, ST2 activation is delayed (35.9 msec) and fires at higher amplitude compared to 7 days. There is notable variability in ST2 firing patterns, as ST2 duration was on average +33.6 ± 46.13% longer at 21 days (Fig. 6). In low (BBB = 16), but not high performing animals, ST2 activation occurs with knee flexion instead of extension during yield (Fig. 5). To determine whether differences in ST2 duration were linear with recovery, burst durations were normalized (percent change postinjury) and correlated with open field BBB scores. A high correlation between ST2 burst duration and BBB scores (r2 = 0.9697; P < 0.05) indicates that smaller changes in burst duration occur in high-performing animals (Fig. 8).
Average burst duration relative to stance onset. Burst durations were measured relative to stance onset (“0”) and averaged before (solid) and 21 days (hatched) after injury. Average EMG onset and offset times are marked by the beginning...
ST2 burst duration predicts recovery in the open field. Normalized burst durations were correlated with BBB scores ranging from 15 to 19. TA, LG, and ST1 display shortened burst durations relative to normal that do not correlate with open field performance...
Changes in ST reflect task specificity
To determine whether different forms of TM locomotion alter muscle recruitment after SCI, we compared flat or 10% downslope grade TM walking in the same animals. Similar to 7 days and 21 days, flat TM walking at 13 days showed delayed activation of ST1 and shorter-burst durations relative to normal. During flat walking, a single prolonged burst with an indiscriminate reset period occurs in ST and ST2 is negligible (Fig. 9). TM walking at a downslope grade required a different recruitment pattern that was identified by changes in the ST. Downslope walking produced later, and less activation of TA for ankle dorsiflexion and recruitment of LG was unchanged (data not shown). In the ST, downslope walking re-established a dual-burst pattern (Fig. 9). Notably, ST2 fired at a greater amplitude with a more defined onset/offset period during downslope walking than flat TM walking (Fig. 9). While downslope walking produced a reset period between ST1 and ST2 within the time period described for Naives, the muscle was not silent.
Task-specific changes in locomotion alter ST recruitment after mild SCI. EMG recordings are shown for the same animal as Naive, and 13 days after injury while walking on flat or 10% downhill TM surface grades. Stick figure diagrams at 60 Hz show a representative...
Overview of the current study
The current work identifies fundamental components of locomotor control that are impaired after recovery from SCI. Despite rapid improvements acutely after injury, deficits persist and normal locomotion does not return by chronic periods. Plateaued walking behavior was characterized by kinematic impairments in yield depicted by a significant decline in angular excursion during the eccentric period of stance (E1–E2). Walking patterns were further characterized by changes in HL muscle recruitment. Delays in activation of knee and ankle muscles occurred during all phases of locomotion. Eccentric actions of the ST (ST2) were notably impaired during yield and significantly correlated with gross open field recovery. Moreover, we found that ST2 activation responds to downslope TM walking after SCI. Our work suggests that the temporal profile of ST serves as a sensitive indicator of gross recovery and that simple changes in locomotor specificity restore its activity.
Changes in neuromotor control after mild SCI
Contusive SCI produces distinct neuropathology with a central core lesion and a peripheral rim of spared white matter that replicates clinical SCI (Bunge et al. 1993; Stokes and Jakeman 2002; Fig. 10). Even with partial sparing of ascending and descending systems, the complex cellular sequellae prevents complete locomotor recovery (Basso 2000; Weaver et al. 2002; Detloff et al. 2008; Fig. 11). Previously, we showed that toe dragging, trunk instability, and paw rotation was associated with white matter sparing between 25 and 60% (Kloos et al. 2005). Here, mild contusion with 34–65% sparing not only produced these persistent deficits during open field locomotion but also significant changes in TM kinematics. The new walking pattern included a more caudal limb position during all phases of gait. As a result, joint angular excursions increased from late stance to mid swing (E3, F phases) but decreased from late swing into yield (E1, E2 phases). Thus, it appears that greater excursion is required to overcome the caudal shift in limb position during the propulsive phases of the step cycle. Unlike the cat, greater flexion was not associated with hypermetria as the paw height during swing was normal (Basso et al. 1994; Fig. 2). Interestingly, the locomotor phases with prominent joint deceleration and lengthening contractions had below normal excursions. This reduction in kinematics during E1 and E2 may be due in part to aberrant motor control strategies. Indeed, alterations in fine control of intralimb coordination are prominent during E1 and E2 phases for both proximal and distal joints (Fig. 4). Moreover, a prevalent, almost uniform delay in neural recruitment of distal HL muscles occurred for the TA, LG, and ST1 (+37.1%, +41.04%, +45.1%, respectively; Fig. 7). To our knowledge, we are the first to quantify recruitment latencies after experimental SCI in rats. Contrary to other SCI models, we did not observe increased recruitment of erector spinae musculature (data not shown), nor did we find aberrant coactivation between muscle pairs or across adjacent joints (Ballermann et al. 2006).
Mild contusion injury results in a central core lesion and peripheral rim of spared white matter. Image depicts the injury epicenter of an animal with a final BBB score of 18 and 64.9% white matter sparing.
The extent of open field recovery correlates with white matter sparing. Endpoint BBB scores are plotted against the percentage of spared white matter (r2 = 0.8502; P < 0.01). Significance determined using Pearson's correlation analysis.
Eccentric motor control is impaired after SCI
Eccentric motor control is a complex skill that emerges late in development (Enoka 1996). During an eccentric contraction, the CNS regulates motor neuron activation to produce muscle forces below an external load resulting in active lengthening. Thus, each lengthening contraction represents the integration of afferent input regarding load and stretch with descending recruitment of motor neurons. Precise CNS modulation prevents muscle spindle-induced stretch reflexes from triggering uncontrolled spasticity after SCI. Other benefits of eccentric contractions include priming the contralateral limb for increased force production, reduced fatigue, and increased metabolic efficiency (Enoka 1996; Grabiner and Owings 1999; Lindstedt et al. 2001). While eccentric actions occur in various parts of the step cycle, the clearest and most predominant occurrence is during weight acceptance or yield phase (E2) when ST and other hamstring muscles lengthen to dissipate impact forces. Our finding that eccentric excursion during yield is markedly impaired across the knee and ankle after contusion confirms previous findings in cats with hemisection SCI (Helgren and Goldberger 1993; Basso et al. 1994; Fig. 4.). It appears that eccentric control of weight acceptance is negatively impacted after SCI and little recovery occurs regardless of injury mechanism or severity.
The ST has a distinct eccentric period of activation that helps determine central pattern generator (CPG)-directed locomotion. Activity in the ST reflects the integration of descending motor drive and afferent input from the limb (Pratt et al. 1996). Phasic sensory signals provided by the second, eccentric burst (ST2) appear to be most important given that it is completely abolished by deafferentation in decerebrate cats and is absent in fictive locomotion unless excitatory drugs are applied (Grillner and Zangger 1984; Grillner and Wallen 1985; Pearson 2004). The magnitude of ST2 activation relates to the rate of knee extension, which suggests that stretch sensitive receptors in ST provide afferent signals to CPGs for locomotion (Wisleder et al. 1990).
We show that recruitment of ST changes over time with recovery. In acute stages, the dual-burst pattern in ST is absent (Fig. 6). A lack in reset between ST1 and ST2 presents a major challenge for a transition to eccentric deceleration in preparation for ground contact. This loss may explain why stepping is not consistent at 7 days. The reset between bursts re-emerges alongside greater activation of ST2 by plateau, but normal patterns are not restored. Interestingly, burst onset and duration of ST2 was the most variable between animals (Fig. 8). Moreover, ST2 activation fails to initiate knee extension before ground contact in low, but not high performing animals (Fig. 5). Thus, it is possible that the integrative function of ST improves with recovery. To determine whether changes in ST were linear with recovery, we compared burst durations of all muscles against open field performance. We found a striking correlation between ST2 duration and BBB scores (Fig. 8). Walking patterns with refined burst duration and a re-established reset period between ST1 and ST2 occurred in animals with greater recovery in the open field. Our work suggests that the temporal profile of ST2 provides a sensitive indication of the spared motor control after SCI. Activity in ST likely reflects the successful integration of spared descending and afferent-driven signals. Facilitating sensorimotor integration in ST may optimize recovery.
Targeted changes in locomotor specificity restore eccentric control after SCI
Activity in ST reflects task-specific changes in locomotion. In the cat, Buford and colleagues show that recruitment of ST changes between forward and backward walking (Buford et al. 1990; Buford and Smith 1990). Similar to our findings early after SCI (Fig. 6), backwards walking eliminated dual bursting and instead elicited a prolonged single burst. The author suggests that the single ST burst may reflect a generic pattern that is modulated by afferent input to produce a double-burst pattern typical in normal locomotion. Clearly, our findings after SCI suggest a lack of supraspinal control across lumbar CPGs. Whether changes in locomotor specificity facilitate activation across lumbar centers after SCI remains unexplored.
Eccentric actions of the ST are accentuated by changing the grade of the TM belt. Steeper grades of downhill TM walking generate progressively greater activation in both bursts of the ST (Smith et al. 1998). After SCI, we find that downslope walking restores a previously dormant ST2 burst (Fig. 9). In early stages of recovery, we show that flat TM walking produces a single prolonged burst in the ST. By tilting the TM belt to a downslope grade, the same animal at the same point in time produces a completely new motor pattern. Indeed, downslope walking restored a reset period and produced greater and more defined activation of ST2. Thus, the rat retained the capacity to produce controlled ST activation in a task-specific manner. This effect may not be observed after more severe lesions, as feline models show an inability to modulate amplitude with slope changes (Brustein and Rossignol 1998).
Conclusions, Limitations, and Future Directions
This study identifies essential features of motor control that do not recover after SCI. Impaired eccentric activity during yield is made evident by changes in kinematics and muscle recruitment. Activity in the ST plays a unique role in locomotor integration and reflects task specificity. Here, we show that impaired actions in ST occur with deficits in yield. Furthermore, we show that improvements in ST functionality indicate the extent of recovery. Whether residual impairments may be resolved after SCI by employing targeted tasks that accentuate eccentric control remains unexplored and warrants further investigation. Changes in locomotor specificity would provide a simple adaptation for current clinical practice.
A limitation to our study is that we could not measure relative amplitude of EMG patterns. Because electrodes were implanted to a chronic time period, we expected exact measurements to be unreliable. In same day recordings (i.e., Fig. 9), interpretations of amplitude are more reliable.
Support for this work was contributed by NINDS#NS07-4882-01A1 (DMB), P30-NS04758, HHSN271200800-0363C (CBSCR).
Conflict of Interest
- Ballermann M, Tse ADY, Misiaszek JE, Fouad K. Adaptations in the walking pattern of spinal cord injured rats. J. Neurotrauma. 2006;23:897–907.[PubMed]
- Basso DM. Neuroanatomical substrates of functional recovery after experimental spinal cord injury: implications of basic science research for human spinal cord injury. Phys. Ther. 2000;80:808–817.[PubMed]
- Basso DM, Murray M, Goldberger ME. Differential recovery of bipedal and overground locomotion following complete spinal cord hemisection in cats. Restor. Neurol. Neurosci. 1994;7:95–110.[PubMed]
- Basso DM, Beattie MS, Bresnahan JC. A sensitive and reliable locomotor rating scale for open field testing in rats. J. Neurotrauma. 1995;12:1–21.[PubMed]
- Basso DM, Beattie MS, Bresnahan JC. Descending systems contributing to locomotor recovery after mild or moderate spinal cord injury in rats: experimental evidence and a review of literature. Restor. Neurol. Neurosci. 2002;20:189–218.[PubMed]
- Behrmann DL, Bresnahan JC, Beattie MS, Shah BR. Spinal cord injury produced by consistent mechanical displacement of the cord in rats: behavioral and histologic analysis. J. Neurotrauma. 1992;9:197–217.[PubMed]
- Brustein E, Rossignol S. Recovery of locomotion after ventral and ventrolateral spinal lesions in the cat. I. Deficits and adaptive mechanisms. J. Neurophysiol. 1998;80:1245–1267.[PubMed]
- Buford JA, Smith JL. Adaptive control for backward quadrupedal walking. II. Hindlimb muscle synergies. J. Neurophysiol. 1990;64:756–766.[PubMed]
- Buford JA, Zernicke RF, Smith JL. Adaptive control for backwards quadrapedal walking. I. Posture and hindlimb kinematics. J. Neurophysiol. 1990;64:745–755.[PubMed]
- Bunge RP, Puckett WR, Becerra JL, Marcillo A, Quencer RM. Observations on the pathology of human spinal cord injury. A review and classification of 22 new cases with details from a case of chronic cord compression with extensive focal demyelination. Adv. Neurol. 1993;59:75–89.[PubMed]
- Damiano DL, Martellotta TL, Quinlivan JM, Abel MF. Deficits in eccentric versus concentric torque in children with spastic cerebral palsy. Med. Sci. Sports Exerc. 2001;33:117–122.[PubMed]
- Detloff MR, Fisher LC, McGaughy V, Longbrake EE, Popovich PG, Basso DM. Remote activation of microglia and pro-inflammatory cytokines predict the onset and severity of below-level neuropathic pain after spinal cord injury in rats. Exp. Neurol. 2008;212:337–347.[PMC free article][PubMed]
- Dibble LE, Hale TF, Marcus RL, Gerber JP, LaStayo PC. High intensity eccentric resistance training decreases bradykinesia and improves quality of life in persons with Parkinson's disease: a preliminary study. Parkinsonism Relat. Disord. 2009;10:752–757.[PubMed]
- Enoka RM. Eccentric contractions require unique activation strategies by the nervous system. J. Appl. Physiol. 1996;81:2339–2346.[PubMed]
- Fehlings MG, Tator CH. The relationships among the severity of spinal cord injury, residual neurological function, axon counts, and counts of retrogradely labeled neurons after experimental spinal cord injury. Exp. Neurol. 1995;132:220–228.[PubMed]
- Gillis GB, Biewener AA. Hindlimb muscle function in relation to speed and gait: in vivo patterns of strain and activation in a hip and knee extensor of the rat (Rattus norvegicus. J. Exp. Biol. 2001;204:2717–2731.[PubMed]
- Goslow GE, Jr, Reinking RM, Stuart DG. The cat step cycle: hind limb joint angles and muscle lengths during unrestrained locomotion. J. Morphol. 1973;141:1–41.[PubMed]
- Grabiner MD, Owings TM. Effects of eccentrically and concentrically induced unilateral fatigue on the involved and uninvolved limbs. J. Electromyogr. Kinesiol. 1999;9:185–189.[PubMed]
- Grillner S, Wallen P. Central pattern generators for locomotion, with special reference to vertebrates. Annu. Rev. Neurosci. 1985;8:233–261.[PubMed]
- Grillner S, Zangger P. The effect of dorsal root transection on the efferent motor pattern in the cat's hindlimb during locomotion. Acta Physiol. Scand. 1984;120:393–405.[PubMed]
- Gruner JA, Altman J, Spivack N. Effects of arrested cerebellar development on locomotion in the rat: cinematographic and electromyographic analysis. Exp. Brain Res. 1980;40:361–373.[PubMed]
- Helgren ME, Goldberger ME. The recovery of postural reflexes and locomotion following low thoracic hemisection in adult cats involves compensation by undamaged primary afferent pathways. Exp. Neurol. 1993;123:17–34.[PubMed]
- Johnson WL, Jindrich DL, Roy RR, Edgerton VR. Quantitative metrics of spinal cord injury recovery in the rat using motion capture, electromyography and ground reaction force measurement. J. Neurosci. Methods. 2012;1:65–72.[PubMed]
- Kaegi S, Schwab ME, Dietz V, Fouad K. Electromyographic activity associated with spontaneous functional recovery after spinal cord injury in rats. Eur. J. Neurosci. 2002;16:249–258.[PubMed]
- Kloos AD, Fisher LC, Detloff MR, Hassenzahl DL, Basso DM. Stepwise motor and all-or-none sensory recovery is associated with nonlinear sparing after incremental spinal cord injury in rats. Exp. Neurol. 2005;191:251–265.[PubMed]
- Lindstedt SL, LaStayo PC, Reich TE. When active muscles lengthen: properties and consequences of eccentric contractions. News Physiol. Sci. 2001;16:256–261.[PubMed]
- Pearson KG. Generating the walking gait: role of sensory feedback. Prog. Brain Res. 2004;143:123–129.[PubMed]
- Pratt CA, Buford JA, Smith JL. Adaptive control for backward quadrupedal walking. V. Mutable activation of bifunctional thigh muscles. J. Neurophysiol. 1996;75:832–842.[PubMed]
- Profyris C, Cheema SS, Zang D, Azari MF, Boyle K, Petratos S. Degenerative and regenerative mechanisms governing spinal cord injury. Neurobiol. Dis. 2004;15:415–436.[PubMed]
- Schucht P, Raineteau O, Schwab ME, Fouad K. Anatomical correlates of locomotor recovery following dorsal and ventral lesions of the rat spinal cord. Exp. Neurol. 2002;176:143–153.[PubMed]
- Smith JL, Carlson-Kuhta P, Trank TV. Forms of forward quadrupedal locomotion. III. A comparison of posture, hindlimb kinematics, and motor patterns for downslope and level walking. J. Neurophysiol. 1998;79:1702–1716.[PubMed]
- Stokes BT, Jakeman LB. Experimental modelling of human spinal cord injury: a model that crosses the species barrier and mimics the spectrum of human cytopathology. Spinal Cord. 2002;40:101–109.[PubMed]
- Stokes BT, Noyes DH, Behrmann DL. An electromechanical spinal injury technique with dynamic sensitivity. J. Neurotrauma. 1992;9:187–195.[PubMed]
- Thota AK, Watson SC, Knapp E, Thompson B, Jung R. Neuromechanical control of locomotion in the rat. J. Neurotrauma. 2005;22:442–465.[PubMed]
- Weaver LC, Marsh DR, Gris D, Meakin SO, Dekaban GA. Central mechanisms for autonomic dysreflexia after spinal cord injury. Prog. Brain Res. 2002;137:83–95.[PubMed]
- Wisleder RF, Zernicke RF, Smith JL. Speed-related changes in hindlimb intersegmental dynamics during the swing phase of cat locomotion. Exp. Brain Res. 1990;79:651–660.[PubMed]
Home visits for older adults aim to prevent cognitive and functional impairment, thus reducing institutionalization and mortality. Visitors may provide information, investigate untreated problems, encourage medication compliance, and provide referrals to services.
Methods and Findings
Data Sources: Ten databases including CENTRAL and Medline searched through December 2012. Study Selection: Randomized controlled trials enrolling community-dwelling persons without dementia aged over 65 years. Interventions included visits at home by a health or social care professional that were not related to hospital discharge. Data Extraction and Synthesis: Two authors independently extracted data. Outcomes were pooled using random effects. Main Outcomes and Measures: Mortality, institutionalization, hospitalization, falls, injuries, physical functioning, cognitive functioning, quality of life, and psychiatric illness.
Sixty-four studies with 28642 participants were included. Home visits were not associated with absolute reductions in mortality at longest follow-up, but some programs may have small relative effects (relative risk = 0.93 [0.87 to 0.99]; absolute risk = 0.00 [−0.01 to 0.00]). There was moderate quality evidence of no overall effect on the number of people institutionalized (RR = 1.02 [0.88 to 1.18]) or hospitalized (RR = 0.96 [0.91 to 1.01]). There was high quality evidence for number of people who fell, which is consistent with no effect or a small effect (odds ratio = 0.86 [0.73 to 1.01]), but there was no evidence that these interventions increased independent living. There was low and very low quality evidence of effects for quality of life (standardised mean difference = −0.06 [−0.11 to −0.01]) and physical functioning (SMD = −0.10 [−0.17 to −0.03]) respectively, but these may not be clinically important.
Home visiting is not consistently associated with differences in mortality or independent living, and investigations of heterogeneity did not identify any programs that are associated with consistent benefits. Due to poor reporting of intervention components and delivery, we cannot exclude the possibility that some programs may be effective.
Citation: Mayo-Wilson E, Grant S, Burton J, Parsons A, Underhill K, Montgomery P (2014) Preventive Home Visits for Mortality, Morbidity, and Institutionalization in Older Adults: A Systematic Review and Meta-Analysis. PLoS ONE 9(3): e89257. https://doi.org/10.1371/journal.pone.0089257
Editor: Hemachandra Reddy, Oregon Health & Science University, United States of America
Received: November 10, 2013; Accepted: January 16, 2014; Published: March 12, 2014
Copyright: © 2014 Mayo-Wilson et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was sponsored by a grant from the Nordic Campbell Centre (Denmark). The authors received internal support from the Centre for Evidence Based Intervention, Department of Social Policy and Intervention, University of Oxford (UK). EMW received internal support from the Centre for Outcomes Research and Effectiveness (CORE), Research Department of Clinical, Educational & Health Psychology, University College London (UK). SG is supported by a linked Clarendon Fund & Green Templeton College Annual Fund Scholarship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
About 13% of Americans and 15–20% of Europeans are over 65 years old , and most wish to remain in their own homes for as long as possible , . However, functional decline increases with age and can lead to loss of independence and early death .
Preventive home visits by health professionals aim to increase autonomy through primary, secondary, and tertiary prevention activities. These differ from home care interventions to treat or to rehabilitate people with medical problems. Some programs focus on one risk factor, such as falls. Others include multidimensional geriatric assessment (MGA) to evaluate and to improve medical, functional, psychosocial, and environmental problems and resources. Whilst proximal outcomes differ, these interventions all ultimately aim to prevent negative long-term outcomes like institutionalization and mortality.
Several previous reviews have argued that home visits are associated with clinical and economic benefits –; however, authors have questioned their active ingredients , suggested that resource-demanding processes be replaced with more efficient services , and argued that they should be discontinued altogether .
The United States Preventive Services Task Force does not recommend multifactorial risk assessment for all community-dwelling elderly due to uncertainty of evidence, but this position may be revised as more information becomes available . As previous reviews about home visits are now outdated , an updated synthesis of relevant studies is required to inform guidelines and ongoing research.
We conducted a systematic review and meta-analysis of randomized controlled trials to assess the effectiveness of preventive home visits for community-dwelling older adults (65+ years) without dementia. Through pre-specified subgroup analysis, we also investigated factors that may moderate these effects . Studies that evaluated follow-up home visits directly related to recent hospital discharge (e.g., to assess or to attend a recently treated condition), and studies in which more than 50% of participants had dementia, were excluded.
In December 2012, we searched the following databases from inception and without language restriction: British Nursing Index and Archive, C2-SPECTR, CINAHL, CENTRAL, EMBASE, IBSS, Medline, Nursing Full Text Plus, PsycINFO, and Sociological Abstracts (Text S1). Reference lists from previous reviews and from included studies were examined, and trial authors were contacted for unpublished studies and outcomes.
We analyzed effects on mortality, institutionalization, hospitalization, falls, injuries, physical functioning, cognitive functioning, quality of life, and psychiatric illness. To limit the effects of reporting bias, studies were included based on the characteristics of the participants and interventions rather than the outcomes included in published reports.
Following methods recommended by the Campbell Collaboration and the Cochrane Collaborations , two authors independently reviewed all citations and extracted data from included studies, such as: context, recruitment strategy, study inclusion criteria, demographics, content and delivery of the intervention, and outcomes. We assessed each study using the Cochrane Risk of Bias Tool; risk of bias was judged ‘high’ for blinding of personnel and blinding of participants per se.
Authors of included studies were contacted to supply any unreported information and to provide information to permit intention-to-treat analyses. Where possible, dichotomous data were entered directly into Review Manager (RevMan) Version 5.2 , and relative risks or rate ratios and 95% confidence intervals (CIs) were calculated using Mantel-Haenszel methods. For dichotomous outcomes that were fully reported in all studies, we also calculated the absolute risk difference. Standardized mean differences (SMDs) and 95% CIs were calculated for continuous measures using Hedges g and combined using inverse variance methods. When studies reported more than one measure of a particular outcome (e.g., psychiatric illness measured using two scales), we averaged the results in Comprehensive Meta-Analysis (CMA) Version 2  before entering data in RevMan 5.2. To estimate event rates in studies reporting the number of events but not reporting time at risk, we assumed that (i) all completers were included for the full duration of the study and (ii) dropouts were at risk for 50% of the year in which they died or left the study. Random-effects models were used due to variability in populations and intervention characteristics across studies. In all forest plots, displays extend beyond the range of probable effects (75% reduction to 400% increase; 4 standard deviations difference in means), and studies are ordered by weight.
For critical outcomes included in the Summary of Findings Table (Table S1), we conducted trim-and-fill analyses  to investigate the possibility of reporting bias. Overall confidence in the results was assessed using the Grading of Recommendations Assessment, Development and Evaluation Working Group (GRADE) criteria , .
Investigation of heterogeneity
Differences among included studies were assessed in terms of their participants, interventions, outcomes, and methods. For each meta-analysis, we also visually inspected forest plots to see if the confidence intervals of individual studies had poor overlap, conducted a Chi2 test, and calculated the I2 statistic. We considered meta-analyses to have important heterogeneity when the p value for Chi2 was less than 0.10 and I2 was greater than 25%.
The following subgroups were analyzed when 10 or more studies were available:
- Number of visits (1; 2 to 4; 5 or more);
- Visitor's professional group
- Participant age (≤70, 71–75, 76–80, 81–85, >85);
- Intervention components:
- Falls only (interventions that exclusively targeted falls prevention, e.g., exercise to improve balance and strength);
- MGA (a systematic evaluation of at least 3 of these domains—medical, functional, psychosocial, or environmental);
- Both falls prevention and MGA;
- Neither falls prevention nor MGA.
Meta-regressions were conducted in STATA  for key outcomes (mortality, institutionalization, falls, and functioning) and four moderators (number of visits, participant age, risk of mortality in the control group, and percentage female) using restricted maximum likelihood.
Results of the search
We identified 18784 records, and full texts were obtained for 176 records (Figure S1). Thirty papers were secondary reports of a study reported in another paper; thus, 146 studies were assessed for eligibility. Post-hoc, we included two studies (both identified in the search) in which participants were assigned using quasi-random methods that approximated the characteristics of randomization as described below , . Sixty-four studies reported in 86 citations were included in the narrative synthesis (Table S2) –, but three of these did not report any outcomes that could be included in a meta-analysis , , .
Seventy-six studies (84 citations) – were excluded for reasons that are enumerated (Table S3). We also identified four ongoing studies – and two studies that could not be obtained , .
Description of studies
Overall, the 64 included studies assigned 28642 participants, ranging from 59 ,  to 3743 , with a median sample size of 299 per study. Follow-up periods ranged from 3 months  to 60 months .
Studies took place between 1981  and 2012  in 13 countries, but most were conducted in the USA (14), UK (14), or Canada (11). Studies used varying eligibility criteria; some included people at high risk of institutionalization while others recruited from the general population. Between 0% and 33% of control subjects died before the last follow-up. Participants were recruited through primary care providers (24), general population registries (11), community and social service organizations (7), emergency rooms (6), health insurance plan registers (5), advertisements (4), veterans' health organizations (1), and various combinations of the above (3); 3 studies did not report how participants were recruited. Studies included participants aged over 65 years (1), 70 years (10), 75 years (28), 80 years (18), and 85 years (3). In others (4), the mean age was over 70 years, but some participants could have been under 65 years. One of these studies included people aged 17 to 99 years; the mean age was 69 years, and 75% of participants were over 65 years .
There was heterogeneity across studies in the number, duration, and focus of visits. Participants received an average of 4.9 (SD = 4.55) visits per study. The number of visits varied by participant in 8 studies. Eleven studies provided one visit per participant, and one study averaged 30 visits per participant. Visitors were nurses alone (27); other professionals, including health visitors, physiotherapists, social workers, and occupational therapists (20); or a combination of health professionals, usually a nurse in combination with another professional (17). Visits had different but overlapping goals, including falls prevention (17), multi-dimensional geriatric assessment (25), both of the above (16), or another prevention (6). Exercise was included in 21 studies. Many studies did not systematically report specific aspects of program design, components that staff actually delivered, or participant take-up.
Comparisons included usual care (50), attention-matched controls including social visits (10), and wait-lists (3); 1 study did not describe the comparison condition. We would have considered comparisons separately, but we could not determine reliably what the comparison groups actually received across different locations, times, and service settings.
Quality of the evidence
Most studies adequately described randomization and were at low risk of bias (41), but sequence generation was unclear in 20 (Figure S2). Allocation concealment was also adequate in 33 studies at low risk of bias, but unclear in 27. Two quasi-random studies were included post-hoc and were rated high risk for randomization and allocation concealment per se, , but we concluded that the methods of assignment had the desirable characteristics of randomization. One study replaced a few intervention participants and was rated high risk for sequence generation and allocation concealment .
Many studies did not describe what happened to participants living in the same household (e.g., husband and wife) and may have randomized small clusters. No study reported that effects were adjusted for clustering; however, studies that explicitly assigned households had cluster sizes close to one.
Subjective outcomes were at high risk of bias for provider and participant blinding; however, mortality, institutionalization, and hospitalization are not likely to have been affected by biased reporting or assessment. Outcome assessors were not blind in 12 studies, which were at high risk of bias for some outcomes, and it was unclear if assessors were blind in 6 studies; other studies were at low risk of bias.
Missing data were unlikely to affect estimates of effects for dichotomous outcomes, including mortality, institutionalization, and hospitalization. As a result, 31 studies were at low risk of bias for incomplete outcome data, 16 studies were unclear, and 17 were at high risk of bias, including two that excluded participants from analyses if they refused visits or did not comply with the protocol , .
Risk of selective outcome reporting was unclear in 43 studies that did not reference a protocol, and there was high risk of bias in 18 studies that omitted measured outcomes. Only 3 studies were clearly free of selective outcome reporting (i.e., outcomes were registered and reported in full).
In trim-and-fill analyses (Figure S3), there was some evidence of small study bias—studies were trimmed for mortality (3), institutionalization (1), falls (2), and hospitalization (5)—but there was little evidence of benefits, and the adjusted effects were not importantly different from the observed effects.
Fifty-three studies (83%) with 23826 participants (83%) reported effects for all-cause mortality. There was high quality evidence of a small relative effect (weighted average) at longest follow-up (Risk ratio = 0.93 [0.87, 0.99]; Chi2 = 54.89, df = 53, p = 0.40; I2 = 3%), but the absolute difference in mortality was close to zero and unlikely to be clinically important (risk difference = 0.00 [−0.01, 0.00]; Chi2 = 64.72, df = 55, p = 0.17; I2 = 15%). Effects for specified follow-up periods were similar to the effect at longest follow-up (Figure 1). Meta-regression did not identify any effects for age (Figure S4) or number of visits (Figure 2); there was a significant relationship with baseline risk of mortality (i.e. annualized risk in the control group), but the relationship was small and unimportant for most studies within the observed range (Figure S4). There was no difference among subgroups when we compared studies by focus of intervention, average age, or number of visits; however, the effect for interventions including both MGA and falls prevention was larger than the effect for either alone. There was some heterogeneity across types of visitors; there was no overall benefit of interventions delivered exclusively by nurses (Figure S5). The data are available as a RevMan 5.2 file (Dataset S1).
Figure 2. Mortality meta-regression by number of visits.
Circles represent studies (N = 47), size represents weight in the analysis. The slope was not significant (0.008 [95% CI −0.02 to 0.04]; t = 0.53, p = 0.60).
Twenty-six studies (41%) with 16264 participants (57%) reported effects for the number of participants in each group who were admitted to an institution. There was moderate quality evidence of no clinically important difference at longest follow-up in relative effects (risk ratio = 1.02 [0.88, 1.18]; Chi2 = 37.64, df = 26, p = 0.07; I2 = 31%) or absolute effects (risk difference = 0.00 [−0.01, 0.01]; Chi2 = 45.13, df = 27, p = 0.02; I2 = 40%). Effects for specified follow-up periods were similar to the effect at longest follow-up (Figure 3). Meta-regression did not identify any effects for age, number of visits, or risk of mortality. There was no evidence of any differences by time point, focus of visit, age of participants, type of visitor, or number of visits (Table 1).
Fifteen studies (23%) including 6288 participants (22%) reported the number of people admitted to hospital in each group (Table 2). At longest follow-up, there was moderate quality evidence of a small relative effect (risk ratio = 0.96 [0.91, 1.01]; Chi2 = 13.70, df = 14, p = 0.47; I2 = 0%) that may not be clinically important (risk difference = −0.01 [−0.03, 0.00]; Chi2 = 13.89, df = 14, p = 0.46; I2 = 0%). There was no evidence of any differences among subgroups, except a statistically significant difference among types of visitors that was consistent with the results for mortality.
Eleven studies (17%) including 4943 participants (17%) reported the number of admissions to hospital. There was low quality evidence at longest follow-up, which would be consistent with no effect or a small effect (rate ratio = 0.93 [0.81, 1.06]; Chi2 = 28.07, df = 11, p = 0.003; I2 = 61%). There was no evidence of any differences among subgroups except a significant difference among types of visitors that was not consistent with the results for people admitted to hospital or mortality.
Twelve studies (19%) including 4321 participants (15%) reported the number of people who visited an emergency room in each group. There was moderate quality evidence at longest follow-up, which would be consistent with no effect or a small relative effect (risk ratio = 0.91 [0.81, 1.03]; Chi2 = 16.29, df = 11, p = 0.13; I2 = 32%). There was no evidence of any differences among subgroups.
Ten studies (16%) including 5870 participants (20%) reported the number of emergency room visits. There was low quality evidence at longest follow-up, which would be consistent with no effect or a small effect (rate ratio = 0.92 [0.81, 1.04]; Chi2 = 35.81, df = 9, p<0.0001; I2 = 75%). Several differences across subgroups were statistically significant because one group in several analyses included only one study that was inconsistent with others; there were no meaningful differences.
Twenty-three studies (36%) including 7455 (26%) participants reported the number of people who fell. One study reported an adjusted effect that could not be combined with other measures to estimate a relative risk, so an overall odds ratio was calculated . There was moderate quality evidence of a small effect at longest follow-up, but it was not statistically significant (odds ratio = 0.86 [0.73, 1.01]; Chi2 = 43.59, df = 22, p = 0.004; I2 = 50%). Most effects were measured after about 12 months; two studies reporting longer follow-up found no evidence of extended benefits. Meta-regression did not identify any effects for age, number of visits, or risk of mortality. There was no evidence of any differences among subgroups, though only one study reported falls but did not explicitly target falls prevention.
Fifteen studies (23%) including 5319 (19%) participants reported the number of falls. There was low quality evidence of a small effect at longest follow-up (rate ratio = 0.74 [0.58, 0.93]), but as with days in hospital or days in institution, the results were extremely inconsistent (Chi2 = 4574.87, df = 14, p<0.00001; I2 = 100%). Some subgroups analyses suggested statistical differences among groups, but studies within these groups were also highly heterogeneous; that is, differences between subgroups did not appear to explain the observed heterogeneity.
Physical and cognitive functioning.
Twenty-seven studies (42%) including 8769 (31%) participants reported a measure of functioning activities of daily living (ADL) or instrumental activities of daily living (IADL). Several studies reported the number of people dependent or independent (or having difficulty) in specific activities (e.g., eating or dressing), but did not report an estimate of overall functioning . There was very low quality evidence of a small effect on ADLs and IADLs at longest follow-up (SMD = −0.10 [−0.17, −0.03]; Chi2 = 55.40, df = 26, p<0.001; I2 = 53%). Meta-regression did not identify any effects for age, number of visits, or risk of mortality. There was no evidence of any differences among subgroups.
Eight studies (13%) including 1608 (6%) of participants reported a measure of cognitive functioning. There was low quality evidence of no clinically important difference at longest follow-up (SMD = −0.06 [−0.21, 0.09]; Chi2 = 12.49, df = 7, p = 0.09; I2 = 44%). We did not compare subgroups due to a lack of studies.
Quality of life.
Twenty-nine studies (45%) including 9892 participants (35%) reported any measure of health-related quality of life. There was low quality evidence of no clinically important difference at longest follow-up (SMD = −0.06 [−0.11, −0.01], Chi2 = 35.69, df = 28, p = 0.15; I2 = 22%). There was no evidence of any significant differences among subgroups.
Psychiatric illness (anxiety and depression).
Fifteen studies (23%) including 3318 participants (12%) reported psychiatric illness (anxiety or depression). There was low quality evidence of a small effect at longest follow-up (SMD = −0.10 [−0.18, 0.02]; Chi2 = 18.06, df = 14, p = 0.20; I2 = 22%). There was no evidence of any differences among subgroups.
Additional analyses identified no evidence of important benefits for: days in institution, days in hospital, fear of falling, people injured, and number of injuries. These outcomes were infrequently reported and many were heterogeneous (Table 2); subgroup analyses did not reveal any patterns that were inconsistent with the results above (Figure S5).
Over the past 20 years, many reviews have investigated the effects of preventive home visiting. Some analyses conclude that comprehensive geriatric assessment may have several benefits , , but other reviews have come to conflicting conclusions , , , , ,  guidelines reflect this uncertainty , but this review finds high quality evidence that preventive home visits do not have important effects on mortality or on independent living overall (Table S1). Future guidelines might recommend against these interventions as they do not have proven effects.
Including 64 randomized trials conducted over the three decades, this review is the most comprehensive in scope, and it identified several studies not included in previous analyses. The results include a wide range of outcomes, the main results are statistically precise with little evidence of statistical heterogeneity, and broad inclusion criteria facilitated several pre-specified subgroup analyses. The quality of evidence was high for mortality and moderate for other critical outcomes, and we conclude that these interventions are ineffective overall; however, we cannot exclude the possibility that some programs may be associated with benefits.
Some home visits are part of larger programs that might have positive effects, including exercise, improved assessment by medical professionals, or falls prevention. However, no specific components appeared to distinguish effective programs from ineffective programs for mortality and institutionalization. Consistent with a recent review of interventions to prevent falls , we find some evidence that home visits may reduce risk of falling; however, many studies did not have a pre-defined primary outcome, and most studies were not prospectively registered. We interpret these results cautiously because other included studies may have measured proximal outcomes and failed to report them. Furthermore, there was no evidence of effects on distal outcomes among those studies that reported reductions in falls.
An overview of reviews identified a need for further analyses to investigate differences related to the focus of visits, number of visits, characteristics of participants, and characteristics of providers . To the extent possible, this review investigated these variables and failed to identify any patterns across outcomes that would be consistent with benefits overall or consistent with benefits for any defined subgroup. Limited reporting of intervention implementation prevented further investigation into potential mediators and moderators (Figure S6). It is possible that some combination of components in particular populations and settings could yield benefits; however, most studies failed to describe a clear program theory . Some interventions included mostly assessment and recommendations; the efficacy of such interventions depends on adherence to these recommendations (e.g., removing fall hazards, exercising, changing medications) and complementary care. Included trials rarely reported participant compliance and use of other services during the trial. As many of the programs required contacts with local health services, it is impossible to assess effects without a description of usual care for each site at each point in time.
Poorly reported trials waste scare resources ,  and can exacerbate difficulties in systematic reviews of complex interventions , , yet under-reporting of intervention content and delivery is common –. This review cannot demonstrate if home visits delivered as intended will have insignificant effects; null effects could be related to implementation failure. To produce useful evidence about the effects of complex interventions, researchers must adhere to CONSORT guidelines for reporting trials , ; report all outcomes measured ; and clearly describe program design, delivery, and uptake so that intervention components can be fully considered in systematic reviews .
Given the size of this review and the number of previous reviews on this topic, further small studies comparing multi-component preventive home visits to usual care will add very little to the knowledge base. Only a very large trial—or a program of research leading to one—could be justified at this time. Funders should not support further trials unless researchers can explain how new studies would change conclusions drawn from a large body of existing evidence.
Home visiting is not consistently associated with differences in mortality or independent living, and investigations of heterogeneity did not identify any subset of programs that are associated with consistent benefits. Due to poor reporting of intervention components and delivery, we cannot exclude the possibility that some programs may be effective. If researchers continue to evaluate these types of interventions, they should begin with clear theories of change, clearly describe programmes and their implementation, and report all outcomes measured , , .
We thank Camilla Thorgaard (Danish National Centre for Social Research) and Karianne Thune Hammerstrom (Norwegian Knowledge Centre for the Health Services) for assistance with the first searches, and G.J. Melendez-Torres (MPhil) at the Centre for Evidence-Based Intervention, University of Oxford, for assistance with generating the meta-regression figures. Thanks also to Terri Pigott (PhD), Jeff Valentine (PhD) from the Campbell Methods Group, and to Nick Huband (PhD), Julia Littell (PhD), Jane Dennis (PhD) from the Campbell Social Welfare Group, and Katherine Hicks (MA) from the Cochrane Dementia and Cognitive Improvement Group, for comments on the protocol and manuscript.
Conceived and designed the experiments: EMW JB KU PM. Performed the experiments: EMW SG JB AP KU PM. Analyzed the data: EMW SG AP. Wrote the manuscript: EMW SG.
- 1. United Nations (2011) World Population Prospects: The 2010 Revision. New York: United Nations Department of Economic and Social Affairs, Population Division.
- 2. Department of Health (2001) National Service Framework for Older People. London: HMSO.
- 3. Rostgaard T, Friedberg T (1998) Caring for Children and Older People - A Comparison of European Policies and Practices. Copenhagen: The Danish National Institute of Social Research.
- 4. Guralnik J, Ferrucci L (2002) Underestimation of disability occurrence in epidemiological studies of older people: Is research on disability still alive? J Am Geriatr Soc 50: 1599–1601.
- 5. Beswick AD, Rees K, Dieppe P, Ayis S, Gooberman-Hill R, et al. (2008) Complex interventions to improve physical function and maintain independent living in elderly people: A systematic review and meta-analysis. Lancet 371: 725–735.
- 6. Elkan R, Kendrick D, Dewey M, Hewitt M, Robinson J, et al. (2001) Effectiveness of home based support for older people: Systematic review and meta-analysis. BMJ 323: 719–725.
- 7. Gillespie LD, Robertson MC, Gillespie WJ, Lamb SE, Gates S, et al. (2009) Interventions for preventing falls in older people living in the community. Cochrane Database Syst Rev 2: CD007146.
- 8. Michael YL, Whitlock EP, Lin JS, Fu R, O'Connor EA, et al. (2010) Primary care–relevant interventions to prevent falling in older adults: A systematic evidence review for the U.S. Preventive Services Task Force. Ann Intern Med 153: 815–825.
- 9. Stuck AE, Egger M, Hammer A, Minder CE, Beck JC (2002) Home visits to prevent nursing home admission and functional decline in elderly people: Systematic review and meta-regression analysis. JAMA 287: 1022–1028.
- 10. Markle-Reid M, Browne G, Weir R, Gafni A, Roberts J, et al. (2006) The effectiveness and efficiency of home-based nursing health promotion for older people: A review of the literature. Med Care Res Rev 63: 531–569.
- 11. Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K (2013) Frailty in elderly people. Lancet 381: 752–762.
- 12. van Haastregt JC, Diederiks JP, van Rossum E, de Witte LP, Crebolder HF (2000) Effects of preventive home visits to elderly people living in the community: Systematic review. BMJ 320: 754–758.
- 13. Moyer VA (2012) on behalf of the U.S. Preventive Services Task Force (2012) Prevention of falls in community-dwelling older adults: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med 157: 1–8.
- 14. Huss A, Stuck AE, Rubenstein LZ, Egger M, Clough-Gorr KM (2008) Multidimensional preventive home visit programs for community-dwelling older adults: A systematic review and meta-analysis of randomized controlled trials. J Gerontol A Biol Sci Med Sci 63A: 298–307.
- 15. Burton J, Montgomery P, Mayo-Wilson E, Underhill K, Thorgaard C (2009) Protocol for a systematic review: Home visits for prevention of impairment and death in elderly people. Campbell Systematic Reviews 5 (Accessed 21 August 2013).
- 16. Higgins J, Green S (2011) Cochrane Handbook for Systematic Reviews of Interventions. Chichester, UK: John Wiley & Sons Ltd.