of individually chosen bed-height adjustments on the low-back stress of nurses.

JP. Effect of individually chosen bed-height adjustments on the low-back stress of nurses. Scand J Work Environ Health 1994;20:427- 34. OBJECTlVK~ - The effects of height-adjustable beds in hospitals on the subsequent prevalence of low back problems among nurses depend on the capacity to reduce low-back stress by bed-height adjust ment. This capacity was investigated in the present study. M ETHODS - Professional nurses performed patient-handling tasks at a standard and an individually chosen bed height. Peak values and time integrals of spinal compression and shear forces were esti mated with dynamic biomechanical modeling. RESULTS - The bed-height adjustment led to lower values of time-integrated compression (average 8.8% lower), peak shear force (average 9.3% lower), and time-integrated shear force (average 18.1% lower). No significance was found for the effect on peak compression, nor for the results for each individual task. This finding can be explained by the minor adjustments made in comparison with the standard height or by the application of different criteria for bed-height adjustment. C ONCLUSIONS - The decreasing time-integrated forces and peak shear force, without a concomitant rise of peak compression, speak in favor of the use of height-adjustable beds in nursing.

Low-back problems are known as a major occupational he alt h problem . Among the groups of workers most adversely affected by low-back pro ble ms is nursing personnel. Several epidemiologic stud ies have fo und relatively hig h prevalences and incidences of low-back problems and worker's compensation claims for nurses (I , 2). Although the exact nat ure of the mech anism ca using low-back prob lems is not clear, it is generally believed that th e mech anical stress on low-b ack structures in association with nursing activities is the main causative factor. T he ha ndling of patients in bed and the tra nsfer of patients to and from their bed s are co nsidered ac tivities that are the most sus cept ibl e to low -back injury (3)(4)(5).
To app ro ach this problem , several strategies have been proposed (6) . One strategy is the selec tion of workers to improve the capacity of nursing personnel to wit hsta nd low-back stress . Another is aime d at redu cing low-back stress by teaching nurses to apply "safe" patient-handling techniques. The effects, however, of bot h strateg ies on the su bsequent prevalence of low-back pro blems ca n be questione d. Some investigators have not man aged to demonstrate measurable effects, neither for selec tio n pro gr ams aimed at the preven tion of back injuries (7) nor for exte nsive education and tra ining programs for nurses ( 1,2).
A third strategy, seemingly more pro sperous, concerns a red uction of physica l stre ss by redesigning the work system. For instance, several mec ha nical devices whic h might be of help in patient-handling tasks have been suggested, for example, pa tient hoists, patient transfer chairs, patient sliding and roller boards, and gait and ambulat io n be lts . All are aim ed at reducing the pushing, pu lling, or lifti ng forces exerted by nurses. Altho ugh the mecha nical advantage for several de vices has been clearly demonstrated (8,9), the use of the eq uipment meets emotio nal barriers. Nu rse s fee l negative abou t the equipme nt, since getting , insta lling, and using equipment usua lly takes much more time than handling the patient manually and pat ie nts can feel insecure when transferred by equipment (10). Another rela tively simple metho d to optimize the work syste m, obvious ly less dram atic for the nur se and the patient, co ncerns the use of beds whi ch are adj ustable in hei ght.
The effect of bed height on spinal fo rce s during the task of turni ng a patient in bed was studied by Gag non et al ( 11). Peak compressive forces on the lumbar spine were found to be 21% lowe r in a "high" bed po sition (at individual hip le vel) co mpa red wit h 427 Scand J Work Environ Health 1994, vol 20, no 6

Tasks
Th e subjects perform ed five patie nt-handling tasks se lected on the basis of a prelimi nary study on the attitude of nurses toward s nur sin g tasks . The five

Subjects
Fou rteen female and eight mal e nur ses participated in the experiments. They were employed in different units of two hospitals in Belgium. Nine of them reported a history of low-back probl ems. At the time of the experiments all of the subjects were free of low -back problems or any other he alth problem. The subje cts' characteristics are listed in table 1. a " low" bed position (0. 10 m below hip level). Thi s difference was attributed to the finding that the trunk was nearer vert ical when a high bed was used , the result being smaller maxim al muscle moments. On the other hand, Lindbeck & Engkvist (12) did not find a difference in L5-S 1 moment s betwe en the use of a low bed (4 1% of body heigh t) and a high bed (46 % of bod y height).
As the optimal bed height probably depends on the task performed and certainly depend s on the individual nurse, the use of beds of an adjustable height migh t be preferred above beds of a fixe d height (whic h are still in use in many nursin g enviro nments). Th e effect of the implementatio n of adj ustable beds in hospitals on the prevalenc e of low-back problems among nurs es is however unknown. An impor tant related question is whether the individual nurse is capable of decreasing the mechanical stress on his or her low back by adjusting the bed height.
In the present study the mechani cal stress on the low back was investigated in different patient-handlin g and transferring tasks. These tasks were performed by profess ional nurses at two bed height s: a standar d fixe d bed height of 0.715 m above ground level [comm on height in the academic hospital (AZVUB) in Brussels] and a bed height chose n by eac h nurse for each task perform ed. Th e mechan ical stress on the low back was ex press ed by peakand time-integrated values of co mpress ive and shea r forces on the lumbar spine, whic h were estim ated on the bas is of a dynamic biom ech anical model. Th e aim of the study was to determine the effec ts of individually chosen bed-h eight adjustments on the various estimates of mechanical low-b ack stress .

Meas urements
Kinematic data, ground reaction forces, and anthropometric data were required to apply the biomechanical model used in the investigati on .
To analyze movements, we used light reflective marker s placed on the subject' s right side at relevant anatom ical positions [the fifth metatarsophalan geal j oint, the distal part of the late ral malle olus, the lateral femo ral epico ndy le, the upperm ost margin of the greater trochanter, the intervertebral disk betwee n the fift h lumb ar and first sac ral vertebra (L5-S I ) from a lateral view (according to reference 14), the spinous process of the first thoracic (T l) verte bra , and the acromion ). Th e instantaneous posi tions of these markers dur ing movement were recorded by use of a direct moti on anal ysis syste m (VICON, Oxford Metrics) at a rate of 60 frames· S-I . Th e marker ' s coordi nates in the sagi ttal plan e of motion were lowpass filtere d (effectiv e cut-off frequency of 5 Hz, zero phase lag, seco nd order, Butterworth). Fro m the filtered body-marker positions a dyn amic linked segment model was co nstruc ted co mprisi ng the feet, the tasks were among the tasks that were experienced as the most stressful (13 ). The tasks largely comp rised movem ents that were lim ited to the sag itta l plane. The subjects were told to perform the tasks at a "normal" steady pace and to appl y the patient-handling techn iques prescribed by a professional nur sing teache r. The man who acte d as the patient weig hed 78 kg, had a height of 1.86 m, and was 25 years of age. He was instru cted to rem ain as relaxe d as po ssible during the trials, hence, not to co-operate nor to work against the nur ses' efforts. The foll owing five tasks were perform ed: (i) turn ing the lyin g patient ove r from his back to his left side, (ii) positioning the lying patient on a bedp an by lifting and turning the patient from his left side to his back, (iii) pulling the lying patient towards the head of the bed , (iv) pullin g and lifting the patient from sitting on the edge of the bed to standing on his feet, (v) lowering the pat ient from standing position to sittin g on the edge of his bed .
Each tas k was performed at two different bed height s, at a fix ed bed height of 0.7 15 m (fro m ground level to the upper side of the mattress), which is a standar d bed hei ght in the academic hosp ital (AZVU B) in Bru ssel s, and at a variable height that was adjusted by the subj ects. The procedu re for adj usting the height was that, first, the bed was lowered to its lowest position (0.30 m above ground level) by the researcher and , next, the nurses were told (without any additional rem ark s or advice from the researc her) to adju st the bed height to their ow n comfort. This procedure was perform ed prior to tasks I and 2 (performed successively in one sessi on) , prior to task 3, and prior to tasks 4 and 5 (also performed in one sess io n). Th e sequence of these sessions was varied sys tematically among the subjec ts. 1 lower legs, the upper legs, the pelvis, and the trunk , which were interconnected at the ankle, the knee, the hip, and the L5-S I joi nt (14). Verti cal and fore-aft components of the ground reaction forces were record ed by means of a force platform (Kistler). Analog force signals were lowpass filtered (30 Hz, 4th order at 24 DB/oct), sampled at 60 Hz. The sampling of the ground reaction force signals and marker coordinates was started and ended by the VICON computer (by one key stroke). From the force components the point of application of the ground reaction force was calcul ated.
Prior to the expe riment s, body height, total body mass, and body segment lengths were measured . On the basis of these measurements and tables (15), the segmental masses, the moments of inertia, and the relati ve position s of the centers of gravity were estimated for each subjec t.

Biomechanical model
On the basis of the kinematic dat a, the measured forces and the anthropometric data, the instantaneous net j oint moment s and joint reacti on forces at the ankle, the knee, the hip, and finalIy at the L5-S I joi nt were estimated by means of an inverse dynamic analysis (16). Next , compressive and shear forces were estimated on the assumption that the L5-S I moment was generated by a single extensor muscle acting 0.06 m posterior to the center of the disk. The possible effec t of the abdominal pressure to support an extendin g lumbar moment , which has been seriously doubted ( 17), was ignored in the calculations.

Data analysis
In each trial the beginning and end points of the task were determined on the basis of the time history of the ground reaction force. Next, peak and time-integrated values for the L5-S I moment s and compressive and shear forces were calculated. For each subjec t the result s from different trials of the same task were averaged. On the basis of an analysis of variance with repeated measure s, the significance was Scand J Work Environ Health 1994, vol 20, no 6 determined of the (co mbined) effects of the type of bed used (adjustable or fixed height) and the tasks performed . In addition, the significa nce of the differences in the results for each handling task between the fixed and adj usted bed height was tested by posthoc comparisons (level of significa nce 0.05).

Results
The adjustments chosen for bed height (with respect to the standard height) are presented in table 2. For the tasks of turning (task I), position ing (task 2), and pulIing (task 3) the patient, most of the subjects chose a higher bed position in comparison with the standard position. For the tasks of lifting (task 4) and lowering (task 5) the patient , the percentages of subjects who preferred a lower and a higher bed position were equal. The mean adj ustments were +65 mm for tasks I and 2, +40 mm for task 3, and +4 mm for tasks 4 and 5, which implied, respectively, bed heights that averaged 46.1 (SD 2.7)% , 44.7 (SD 2.9)%, and 42.5 (SD 3.1)% of the subject's body height. The observed intersubject variation in bed-height adjustment might have been expected beforeh and, when the variation in body height among the subjects was taken into consideration (range 1.57-1.81 m), However, it was also found that the bed height chosen did not correlate with body parameter s like body height (figure 1) or hip height.
The resulting values for the mechanical parameters for the standard and adj usted bed heights are presented in figures 2 and 3. The statistical result s are listed in table 3.
With respect to spin al compr ession, the peak values ranged from 2626. 3 (SD 715 .0) N to 4575.7 (SD 857.2) N for the tasks performed with the standard bed height. A tend ency toward s slightly lower peak values were observed for the adj usted bed height for four of the five tasks (figure 2). However, the effect of bed-height adjustment on peak compression was not significant. On the other hand , a statistically significant effect of bed height adjustment was found for the time integrals of spinal compression.

Discussion
The effect of ins talling height-adju stabl e beds in nursing environments on subsequen t low-back problems depends on the nurse' s capacity to reduce the low-back stress by bed-height adj ustment. The present study demon strat ed significant, fa vorable effects of individual bed-h eight adjustment on the fol-626 .9 (SD 101.5) N. In general, smaller values were found with the use of the adjusted bed height , an average of 35.4 N, or 9.3%, for the pea k values and an average of 108.3 N . S-I , or 18.3%, for the integra ted values. The se general effects of bed height were statistically significant. For the individual tasks, significantly lower values were found for the adjusted bed height position onl y with res pect to the peak shea r force in tas k I ("t urn ing") and the tim e-integra ted shear force in task 3 ("pulling") . Additio na l statistical tests were performed to investigate severa l subject-related factors . Although the male nurses were subjected to higher spina l forces than the fema le nurses (which could be attri buted to their higher total body mass), the effects of bedheight adjustment on the mechanical parameters were similar for both genders . In addition, no sig nific ant differences, with respect to the mechanical stress parameters or the effect of bed-h eight adjustm ent , were found betw een the nur ses with and tho se without a history of low-back problems . In addition, a comparison was made bet ween the nurses who had experience adjusti ng bed positions in their daily work and nurses who had no such experie nce. It was found that the bed-h eight adjustments made by the experienced nurses led to a significant decrease in time-integrated co mpressio n and peak and time-int egrated shear forc e, while the effec ts were not sig nifica nt for the inexperienced nurses. How ever, as the size of the inexperienced group was small (N = 3). the data for the effect of expe rience are too few to be considered care fully.
Finally, the effects on low-back stress were studied in relation to the adju stments made. It was found that tendencies towards lower values for the stress paramet ers with the bed adj ustme nt were more pronounced for the subjec ts who raised the bed than with those who lowered it. When bed height was expressed as the percentage of body height , the stress paramet ers fo r adj usted height tended to be the mos t pronounced for the subjects who chose a bed height at a relatively high percentage of their ind ividual body height (in comparison with the tota l grou p). Although most of the tendencies were not significant (mainly due to the necessity to split the tota l data set into smaller sets, both on a subject and a task level), the se result s sugg est that the extent of favo rab le effects of bed height adju stment was limit ed by subj ects who preferred a relatively low bed height in relation to their individua l body height. Turning   lowing mec hanical parameters: peak and time-integrated shear force and time -integrated spinal compression. On the other hand , there was no significance for the general effect of bed -height adjus tmen t on peak compression, or for differences in the results for the individual tasks between the adjusted and fixed bed height (except for peak shear force in "turning" and integrated shear force in "pulling") . As is discussed later, several factors might have prevented the latter res ult s from being significant.

Factors interfering with fa vorable effects
An obvious factor that cou ld ha ve limited the favorab le effects to some extent is a restricted capacity of the nurses to select an optim al bed position with respect to physical stress on the low back . Although most of the subjects had experience in adjusting bed height for nursing tasks, some might have been unsuc ces sful in red ucing low-back stress simply becau se they were not capable. However, there might have been other interfering factors as well.
In studying the practical advantage of height-adjustable beds, the use of the individually adj usted height was compared with the use of a bed with a standard height (as is generally available) . It is likel y that this stan dard height was already close to optimal for a substantial number of the subjects. This situation would have made it difficult to establish significance for any effect of the adjustment. Moreover, as a result of the supposedly near-optimal standard height, various nurses adju sted the bed to a minor extent in an upward or downward direction, while only a small portion of the nurses raised the bed significantly. Wit h respect to maximal spinal compression, it is solely for these latter subjects that the mechanica l advantage is clear. An increase in bed height yields a trunk position nearer the vertical and resu lts in a lower muscle moment at the lumbar level (I I).
Finally, the adjustments might have been chosen according to criteria other than minimizing low-back stress . Such a criterion might be the limitation of muscular effort in the upper body. It could be hy-432 pothesized that a relatively low bed position mig ht relieve the loading on the arm and shou lder muscles for two reasons. First, as a relatively high bed requires an ere ct standing position (extended legs and trunk) , the power required to handle a patient can onl y be generated by the muscles of the upper extremities . With a relatively low bed, however, a substantial part of the muscle power can be generated by flexin g and extending the lower extremities and trunk. Second , when the trunk can be flex ed at this lower. bed height, the patient can be handled clo ser to the horizontal position of the shoulde rs. This possibility implie s a lower shoulder muscle moment than with the vertically erect trunk position impo sed by a high bed. An attempt to relieve stress on upperextremity muscles might have been the reason for some of the subjects choosing a relat ively low bed posit ion in relation to their body height. As noted befo re, thes e subjects might have limited the favorable effects of bed adjustment on low-back stress to some extent. Thus the application of different criteria by the nurses studied may have formed another obstacle to the achievement of signi ficance for some of the results.

Individually chosen bed height in relation to body height
In the tasks in which the patient was handl ed in bed , the bed height wa s adjusted to a level that corresponded, on the average, with 46 . I(SO 2.7)% (tasks I and 2) and 44 .7 (SO 2.9)% (task 3) of body height. For lifting and lowering the patient (tasks 4 and 5), this percentage was slightly lower: 42 .5 (SO 3.1) %. The values are in close agreement with the findings of Lindbeck & Engkvist (12). In their study, nurses adju sted the bed height for the task of " moving the patient up the bed" by an average of 44 .7(502.5)%.
In the present study we did not find any relationship between bed-height adjustment and body height (or hip height ). This rather surprising find ing may be expl ained by the aforementioned fact ors of intersubject variation in the application of criteria for bed-height adjustment, the minor extent of bed-height adjustment by some nurses, and restrictions on the nurse's capability to adjust the bed correctly.
As the experiments were designed to determine the capacity of nurses to reduce back stress by adjusting the bed in comparison with a standard position, an optimal bed height cannot be determined from the results. This determination would have required a protocol of several imposed bed heights, from which the least stressful could be determined. From the preceding considerations, it can be stated that an optimal bed height is hard to define, since it not only depends on anthropometric dimensions but also on the nurse's physical work capacities and the task to be performed.

Peak versus time-integrated values
Despite the amount of bed height adjustment and the possible application of criteria other than the reduction of low-back stress, significant favorable effects of bed-height adjustment were observed for the time integrals of compression and shear force and for the peak shear force. Especially the reduction in time integrals is noteworthy. This reduction can be partly explained by a decrease in the time required to perform the tasks at the adjusted bed height.
One can only speculate about the relevance of the results with respect to the risks of low-back problems among nurses, for the etiology of low-back problems is still unclear (18). The main interest in ergonomics has always been to reduce the maximal forces operative on lumbar motion segments during peak loading activities. Most attention has been directed towards the peak compressive forces. The peak values of spinal compression may lead to fractures of the vertebral end plate and the underlying cancellous bone (19), which is a frequent type of damage observed in autopsy studies (20). Several guidelines for manual materials handling [eg, the guidelines of the National Institute for Occupational Safety and Health (NIOSH) in the United States] are (partly) based on the criterion of maximal motion segment compression (21). The present study showed that the NIOSH criterion for maximal compression of 3.4 kN was exceeded in four of the five tasks and that the chosen bed-height adjustments had no effect on this parameter.
Although shear forces have been estimated frequently in biomechanical research on low-back loading, shear force criteria are not available in the literature. Several authors have suggested that the point of damage directly on the vertebrae due to shear force is hardly reached in normal human activities (22,23). On the other hand, it is generally believed that shear forces might lead to damage to the neural arch in human activity, as ventrally oriented shear force is counteracted by the facet joints. In addition, the facet joints have been found to be a direct source of pain (24). In the present study, we found that bed-Scand J Work Environ Health 1994, vol 20, no 6 height adjustment leads to an average decrease of 9.3% in peak shear force. Whether this decrease is enough to decrease the risk of low-back problems can only be established on the basis of longitudinal epidemiologic research.
In contrast with the interest in the (peak) magnitude of forces, temporal aspects of the mechanical stress have received very little attention, although these aspects are likely to be highly relevant (25). Because of the visco-elastic nature of biological tissue, prolonged or repetitive loading may result in residual deformation and cumulative fatigue and therefore reduce the threshold stress at which tissues fail. In a recent epidemiologic study the temporal aspects of the load were incorporated successfully. The integral of the compressive force on the L5-S I motion segment over a workday has been demonstrated to have a positive relation to the prevalence of low-back pain (26). In the present study the time integrals for the compression and shear force were found to decrease by 9.3 and 18.8% through the use of bedheight adjustment. In light of the aforementioned considerations a major conclusion of this study is that nurses are capable of adjusting beds in such a way that the total amount of mechanical stress on the L5-S I motion segment over the total duration of patient-handling tasks (in terms of integrated forces) decreases without peak values of stress being increased.

Concluding remarks
In this study, significant favorable effects of bedheight adjustment were observed for the time integrals of compression and shear force and for peak shear force. The observed tendency for lower peak compression values with bed-height adjustment was not significant. This finding may be due to the minimal bed-height adjustments made (standard bed height might be close to the optimal height for various subjects) and the possible application of criteria other than the reduction of low-back stress. Nevertheless, the results of the present study, especially the reduction of exposure to a high level of force on the L5-S I motion segment without a concomitant rise in peak forces, speak in favor of the use of heightadjustable beds in nursing.