Physical Restraint Initiation in Nursing Homes and Subsequent Resident Health

The Gerontologist Copyright 2008 by The Gerontological Society of AmericaVol. 48, No. 4, 442–452

Physical Restraint Initiation in Nursing Homesand Subsequent Resident Health

John Engberg, PhD,1 Nicholas G. Castle, PhD,2 andDaniel McCaffrey, PhD1

Purpose: It is widely believed that physical restraintuse causes mental and physical health decline innursing home residents. Yet few studies exist showingan association between restraint initiation and healthdecline. In this research, we examined whetherphysical restraint initiation is associated with sub-sequent lower physical or mental health. Designand Methods: We used all nursing homes (N=740)in Pennsylvania in 2001, with 12,820 residents. Weused the Minimum Data Set data; Online Survey,Certification and Reporting data; and the AreaResource File as data sources. We restricted oursample to newly admitted nursing home residentswho were not restrained in the first two quarters oftheir residency. We examined which facility andindividual characteristics during those first twoquarters were associated with restraint initiationduring the third quarter. We then examined theassociation of third-quarter restraint initiation withfourth-quarter health outcomes, using regressions thatcontrolled for first- and second-quarter health status aswell as other resident, facility, and market character-istics. The physical health outcomes examined con-sisted of falls, walking dependence, activities of dailyliving (ADLs), pressure ulcers, and contractures. Mentalhealth outcomes examined consisted of cognitiveperformance, depression, and behavior problems.Results: The initiation of restraint use was associatedwith a previous fall (p , .01), psychoactive medicationuse (p , .05), low cognition (p , .01), ADL scores (p ,

.01), and the absence of pressure ulcers (p , .10), aswell as a variety of facility characteristics. Subsequentto restraint initiation, we found an association withlower cognitive performance (p , .01), lower ADL

performance (p , .01), and higher walking depen-dence (p , .01). Implications: We found that anassociation between restraint initiation and subse-quent adverse health consequences exists and issubstantial. Moreover, these results would appear tohave practical as well as statistical significance.

Key Words: Physical restraint, Outcomes,Health, Nursing homes

A physical restraint is ‘‘a device that is attached . . .and cannot be easily removed by the resident whichrestricts freedom of movement and/or normal accessto his/her body’’ (State Operations Manual, 2002,p. 23). Some believe that these restraints causenegative health outcomes for nursing home residents.But in actuality, many of the pernicious effects ofphysical restraint use are not well investigated, andquestions exist as to whether adverse effects of re-straint use exist. Thus, in this study we investigatedthe nexus between the use of physical restraints andthe subsequent lower health status of residents.

Understanding whether physical restraints con-tribute to health problems is important. As othershave pointed out, it is not good clinical practice, oreven ethical, to physically restrain nursing homeresidents (Evans & Strumpf, 1989). Consumergroups such as the National Citizens Coalition forNursing Home Reform have been particularlysuccessful in sensitizing policy makers, the public,and practitioners about the indiscriminate use ofphysical restraints in nursing homes. However, ifrestraints are shown to cause physical or mentalhealth decline, a more powerful rationale for limitingtheir use may develop, helping further reducerestraint use and improving the health and satisfac-tion of residents. With a concomitant reduction inresidents with health problems, nursing homes maythen better provide services to other impairedresidents.

Prior studies have shown that restraint use may beassociated with mental health problems, including

This study was supported in part by a grant from the Agency forHealthcare Research and Quality, 1 R03 HS013983-01A1 (PI: Castle).

Address correspondence to Nicholas G. Castle, PhD, A610 CrabtreeHall, 130 DeSoto Street, School of Public Health, University ofPittsburgh, Pittsburgh, PA 15261. E-mail: [email protected]

1RAND Corporation, Pittsburgh, PA.2School of Public Health, University of Pittsburgh, PA.

442 The Gerontologist

by guest on Decem

ber 29, 2015http://gerontologist.oxfordjournals.org/

Dow

nloaded from

http://gerontologist.oxfordjournals.org/

increased social isolation and decreased cognitivefunction. Restraint use can create social isolation, asboth other residents and staff avoid restrained resi-dents (Lofgren,MacPherson,Granieri,Myllenbeck,&Sprafka, 1989). Social isolation negatively impacts thehealth of elders (Mor et al., 1995). Folmar andWilson(1989) suggested that the least socially active residentsare most likely to be restrained, which furtherattenuates their social performance. In addition,combative residents often become more combativewhen restrained (Marks, 1992). Burton, German,Rovner, and Brant (1992) suggested that restraintsmay contribute to cognitive decline. Moseley (1997)determined that physical restraint use was associatedwith increased resident disorientation and walkingdependence.

Residents are frequently restrained to preventthem from falling (Capezuti, Evans, Strumpf, &Maislin, 1996). However, restrained residents stillexperience falls. For example, Werner, Cohen-Mansfield, Braun, and Marx (1989) cited a residentwho fell four times, even though he was restrainedduring three of these falls. In an empirical analysis,Mion, Frengley, Jakoveic, and Marino (1989)identified more frequent falls (p=.001) in restrainedpatients. Tinetti, Wen-Liang, Marottoli, and Ginter(1991) found serious fall-related injuries in 17% ofrestrained residents as compared to 5% of thosewho remained unrestrained (significant at p , .001).In contrast, Ejaz, Folmar, Kaufmann, Rose, andGoldman (1994) identified a significant increase infalls after the implementation of a restraint reductionprogram. However, a more recent study by Capezutiand associates (1996) found restraint use to neitherincrease nor decrease falls among nursing homeresidents. Levine, Marchello, and Totolos (1995) didnot observe any increase in the number of falls (orinjuries) over a 3-year period, even though theyreduced restraint use from 39% to 4%. It is inter-esting to note that falls are probably the mostresearched health consequence associated with phys-ical restraint use, yet the findings of these studies areequivocal.

It is also worth noting that federal regulationsmandate that to prevent the damaging side effects ofimmobility due to restraint, restrained residentsshould be released, exercised, and repositioned every2 hours (Schnelle, Simmons, & Ory, 1992). How-ever, Schnelle and colleagues (1992) determined thatfacilities often do not follow this mandate. Rather,they found that facilities with restrained residentsalso have poor restraint management practices. Twoother well-known consequences of immobility arepressure ulcers and contractures. Pressure ulcersaffect both the comfort and the medical outcomesof nursing home residents with impaired mobility.Contractures are an abnormal shortening andstiffening of muscle tissue that can decrease therange of motion at a joint. This can produce a changein gait and a decrease in walking velocity—both of

which are major risk factors for falls—and may alsolimit mobility in daily life.

Based on these prior studies, we hypothesized thatthe initiation of physical restraints would be asso-ciated with lower subsequent mental and physicalhealth outcomes. Following these prior studies, weexamined the following mental health outcomes:cognitive performance, depression, and behaviorissues. Also following these prior studies, we exam-ined the following physical health outcomes: falls,activities of daily living (ADLs), pressure ulcers,contractures, and walking dependence.

These previous studies were mostly observationaland used small sample sizes. The most comprehen-sive empirical examination in this area comes fromCastle (2006), who, using 2,000 nursing home resi-dents, found that restraint use was associated withlower cognitive performance, depression, and socialengagement. However, this study only examinedmental health outcomes (Castle, 2006). In addition,this study, and all others in this area, used limitedstatistical tests that could not account for potentialbiases such as the fact that residents who havealready become impaired may be most likely to berestrained. Some of these studies were done at thefacility level and aggregated nonrepresentative sam-ples of residents. Other studies did not account fordifferent sources of resident and facility variation.Thus, in this analysis we used a large sample size andtook into consideration statistical variation at themarket, facility, and individual level in a regressionmodel of outcomes. Most important, we used thelongitudinal information on a sample of unrestrainednew residents and investigated the factors that areassociated with restraint initiation and the sub-sequent health outcomes of individuals who are andare not restrained.

Methods

Data

Three sources of information constituted the pri-mary project database: Minimum Data Set (MDS)data; Online Survey, Certification and Reporting(OSCAR) system data; and the Area Resource File.

The MDS is a summary assessment of nursinghome residents. It was created to measure residents’functional status, health conditions, services re-ceived, demographics, and payer source and hasnearly 400 data elements, including cognitive func-tion, communication/hearing problems, physicalfunctioning, continence, psychosocial well-being,mood state, activity and recreation, disease diag-noses, health conditions, skin conditions, specialtreatments, and medication use. All Medicare- andMedicaid-certified nursing facilities are required touse the MDS on at least three occasions: (a) onadmission, (b) at least annually, and (c) if theresident shows ‘‘significant change.’’ In addition, all

Vol. 48, No. 4, 2008 443

by guest on Decem


Dow

nloaded from


residents are assessed quarterly on a subset of theMDS. In this analysis, we used both the full MDSand subsequent quarterly MDS assessments ofresidents. The MDS data are carefully constructedand reliable (Hawes et al., 1995).

The OSCAR contains facility and aggregatedresident data routinely collected through the nursinghome certification process conducted by state licensureand certification agencies (Harrington, Zimmerman,Karon, Robinson, & Beutel, 2000). Approximately17,000 facilities are included in the OSCAR. OSCARfiles include data for general facility characteristics,including facility ownership, number of beds, and theaverage census. The validity and reliability of manyOSCAR data elements is debatable (Straker, 1999);nevertheless, the facility characteristics used in thisanalysis generally are considered accurate and reliable(Harrington et al., 2000).

The Area Resource File is a publicly available dataset summarizing a large array of census, health, andsocial resource information. It is not a primary datacollection effort but a compilation of data takenfrom sources such as the American HospitalAssociation annual hospital survey, the U.S. Censusof Population and Housing, and the National Centerfor Health Statistics. Extensive details regarding thisdata can be found on the World Wide Web (http://wonder.cdc.gov/wonder/sci_data/census/arf).

Sample Selection

In this analysis we used residents and facilitiesfrom Pennsylvania. We used this state because ofdata availability and because it represents a sampleof convenience. We excluded short-stay residents,hospice residents, and those in a coma. Althoughrestraint use on all residents is important, coma andhospice residents have special issues that are distinctfrom those of the general population and are beyondthe scope of our study. The outcomes we usedcannot be utilized for coma residents and for short-stay residents (less than 3-month length of stay); andfor hospice residents (less than 6-month length ofstay), multiple data points for the outcomes ofinterest are unlikely and resident health status islikely to be more transitional.

A further analytic consideration was the needto specify an index point in time for the initialassessment of mental/physical health status and useof restraints. One conservative way of doing so,which we used, was to examine ‘‘new’’ nursing homeresidents. We could identify residents new to thenursing home from the MDS. We restricted oursample to residents for whom the MDS contained atleast four records approximately 3 months apart, thefirst of which had to be an admission assessment. Weused only one admission per individual to avoidunmeasured correlation among repeat stays by thesame individual.

The requirement that we observe the admissionrecord plus at least three subsequent records spacedapproximately 3 months apart eliminated all indi-viduals who were admitted before 2001, who wereadmitted later than March 2001, or who stayed lessthan 9 months. We also eliminated nonelderly andcomatose patients. This left almost 14,000 residentsin Pennsylvania. We eliminated approximately 8%who were restrained during the first two quarterlyrecords. This left us with an analytic sample of12,820 residents. This number represented approx-imately 20% of the Pennsylvania resident census for2001. Of these residents, 2.3% were restrained in thethird quarter of their stay. This rate is lower than theprevalence rates cited elsewhere in our study becauseit reflects the incidence of restraint initiation ratherthan the overall prevalence. Table 1 gives these andother descriptive statistics.

Dependent and Independent Variables

We examined the restraint process in two steps.In the first step, we modeled who got restrained.In the second step, we modeled the association ofrestraint initiation with subsequent changes in healthoutcomes.

In the first step, the dependent variable wasphysical restraint initiation. Restraint type and useare given in Section P of the MDS, and five categories(types) of restraint use can be recorded: (a) full bedrails on all open sides of bed, (b) other types of siderails used (e.g., half rail, one side), (c) trunk restraint,(d) limb restraint, and (e) chairs prevent rising. Nofurther detail, other than these five categories ofrestraints, is available from the MDS. However, anational advisory committee created these five cate-gories to be both comprehensive and distinct fromone another (Hawes, Vladeck, Morris, Phillips, &Fredeking, 2003).Thus, it is likely that these categoriesare sufficient to capture restraint use. Furthermore,the inclusion of few ‘‘clarifications’’ in the MDS 2.0Training Manual indicates that the industry is able touse these categories with few problems.

We created a single dichotomous measure ofrestraint use that indicated whether trunk, limb, orchair restraints were used. We excluded informationon bed rail restraints. We used the values of thisrestraint use variable in the first and second quarterto define our analytic sample, which consisted ofindividuals who were not restrained in either of thesequarters. The third-quarter value of this variable wasthe dependent variable in the first step of theanalysis, as described more fully in ‘‘StatisticalAnalysis.’’

The physical health outcomes examined consistedof falls, walking dependence, ADLs, pressure ulcers,and contractures. Mental health outcomes examinedconsisted of cognitive status, alterations in mood,and behavior problems. These are shown in Table 1


by guest on Decem


Dow

nloaded from


Table 1. Dependent and Independent Variables

M (SD) Variable Operational Definition

Dependent variablesa

2.662 (1.653) CPS Measure of cognitive performance using five MDS items (comatose status,short-term memory, ability to make decisions, ability to make selfunderstood, and self-performance in eating); higher values indicatelower cognitive functioning.

2.791 (1.502) ADL scale Five items from the MDS (transfer, locomotion, dressing, eating, and toiletuse); higher values indicate lower functioning.

0.872 (1.587) Depression Measure of depression (MDS-DRS) using seven MDS items (resident madenegative statements; persistent anger and irritability with self andothers; expressions of what appear to be unrealistic fears; repetitivehealth complaints; repetitive anxious complaints/concerns[non-health-related]; sad, pained, worried facial expressions; and crying,tearfulness); higher values indicate greater depression.

0.411 (0.855) Behavior issues Measure of resident behavior problems using four MDS items(verbally abusive, physically abusive, socially inappropriate/disruptivebehavior, and resisted assistance in ADLs); higher values indicategreater behavior problems.

0.160 (0.366) Falls Dichotomous variable indicating having fallen in the past 30 days.0.078 (0.267) Pressure ulcers Dichotomous variable indicating presence of any pressure ulcer.0.019 (0.135) Contractures Dichotomous variable indicating presence of any contracture. MDS records

contractures as: 1 ¼ none, 2 ¼ face/neck, 3 ¼ shoulder/elbow,4 ¼ hand/wrist, 5 ¼ hip/knee, and 6 ¼ foot/ankle.

0.745 (0.436) Walking dependence Dichotomous variable indicating assistance needed when walking in room.

Independent variables

0.023 (0.150) Restraint useb Dichotomous variable indicating use of trunk, limb, or chair restraints.0.093 (0.291) Psychoactive medsc Whether psychoactive medications are given or not, including four

categories of drugs (antianxiety, sedative/hypnotic, antipsychotic,or antidepressant).

83.590 (7.415) Aged Standardized age (in years).0.236 (0.425) Maled Male ¼ 1, female ¼ 0.0.080 (0.271) Blackd Racial/ethnic background ¼ 3 (racial/ethnic background as recorded on

the MDS: 1 ¼ American Indian/Alaska native; 2 ¼ Asian/PacificIslander; 3 ¼ Black, not of Hispanic origin; 4 ¼ Hispanic; 5 ¼ White,not of Hispanic origin).

0.010 (0.098) Other minorityd Racial/ethnic background ¼ 1, 2, or 4 (racial/ethnic background asrecorded on the MDS: 1 ¼ American Indian/Alaska native;2 ¼ Asian/Pacific Islander; 3 ¼ Black, not of Hispanic origin;4 ¼ Hispanic; 5 ¼ White, not of Hispanic origin).

0.411 (0.492) Ownershipd For profit or not for profit (for profit ¼ 1, other ¼ 0).0.514 (0.500) Chaind Whether facility is a member of a nursing home chain.0.623 (0.214) Medicaid occupancyd Average daily Medicaid occupancy rate (fraction of total occupancy).193.202 (154.092) Bed sized Number of beds.0.899 (0.087) Occupancy rated Average daily overall occupancy rate.0.051 (0.029) RNs per bedd FTEs of RNs aides per bed.0.095 (0.046) LPNs per bedd FTEs of LPNs per bed.0.297 (0.096) NAs per bedd FTEs of NAs per bed.2.255 (1.063) NA per nursed FTEs of NAs divided by number of LPNs and RNs.1.166 1.248) Competitiond Market competition measured with the Herfindahl index: each facility’s

percentage share of beds in the county divided by the sum of thesquared market shares of all facilities in the county (0–1) calculated fornursing facilities.

0.878 (0.047) County occupancy rated Average occupancy rate for nursing facilities in the market.0.571 (0.160) Nonprofit market shared Proportion of nonprofit facilities in county.0.330 (0.153) Managed cared Managed care penetration in the county.

Notes: N = 12,820. The dependent variables come from the MDS, measured in the fourth quarter post admission. Theindependent variables restraint use, psychoactive meds, age, male, Black, and other minority come from the MDS; the nonprofitmarket share and managed care come from the Area Resource File; and all remaining independent variables come from the OnlineSurvey, Certification and Reporting system. CPS = Cognitive Performance Scale; MDS = Minimum Data Set; ADL = activity ofdaily living; MDS-DRS = MDS–Depression Rating Scale; RN = registered nurse; FTE = full-time equivalent; LPN = licensedpractical nurse; NA= nurse aide.

aAlso used as lagged variables to control for prior health status measured in the first and second quarters.bMeasured in the third quarter post admission.cMeasured in the first and second quarters post admission.dMeasured at admission.

Vol. 48, No. 4, 2008 445

by guest on Decem


Dow

nloaded from


and described in more detail below. As described in‘‘Statistical Analysis,’’ we used the values of theseoutcome measures in the first two quarters to riskadjust, and the values in the fourth quarter were thedependent variable in the second step of the analysis.

The MDS records two measures of falls: (a) fell inpast 30 days, and (b) fell in past 31 to 180 days.Given that restraint use is likely to cause muscledeconditioning and problems with balance, webelieve a fall could occur as a result of restraintinitiation. We based our measure on falls within 30days because the longer time frame would overlapwith the third-quarter measurement. It was impor-tant that we measure falls subsequent to restraintinitiation.

The MDS records whether a resident can walkindependently in his or her room. We created adichotomous variable that indicated walking de-pendence; we coded the variable as 0 if the residentcould walk independently and 1 if the residentneeded supervision or assistance or could not per-form this task.

We used a score measuring ADL status. Wecreated this from five MDS items: transfers, loco-motion, dressing, eating, and toilet use. We gaveresidents dependent in a category a score of 1, andthose independent a score of 0. Thus, we used a6-point scale from 0 to 5, with higher valuesindicating greater ADL impairment.

The MDS records pressure ulcers as follows: 1 =Stage 1, a persistent area of skin redness (withouta break in the skin) that does not disappear whenpressure is relieved; 2 = Stage 2, a partial-thicknessloss of skin layers that presents clinically as anabrasion, blister, or shallow crater; 3 = Stage 3,a full thickness of skin is lost, exposing thesubcutaneous tissue—presents as a deep crater withor without undermining; and 4 = Stage 4, a fullthickness of skin and subcutaneous tissue is lost,exposing muscle and/or bone. We created a dichot-omous variable that indicated the presence of one ormore pressure ulcers at any stage.

The MDS records contractures as follows: (1)none, (2) face/neck, (3) shoulder/elbow, (4) hand/wrist, (5) hip/knee, and,(6) foot/ankle. We set ourdichotomous measure of contractures equal to 1 ifany of these contractures was present.

Depression was measured using the MDS–Depression Rating Scale (Burrows, Morris, Simon,Hirdes, & Phillips, 2000). This scale is based onseven MDS items and achieved a correlation of .69with the Cornell scale and .70 with the HamiltonDepression Rating Scale. This has a range of scoresfrom 0 to 6.

The MDS also includes a behavior problems indexthat has been used successfully by several research-ers. Snowden and associates (1999) determined thatthe MDS index of behavior problems had a .50correlation with physician-observed behavior prob-lems. This has a range of scores from 0 to 3.

For a quantitative assessment of the mental healthstatus of residents we used the Cognitive Perfor-mance Scale. Gruber-Baldini, Zimmerman, Morti-more, and Magaziner (2000) determined that theMDS Cognitive Performance Scale had a .68 corre-lation with the Mini-Mental State Examination. Thishas a range of scores from 0 to 6.

Control Variables

Outcomes can rarely be fully attributable toantecedent care and are therefore often ‘‘adjusted’’for differences in risk for adverse outcome amongpatients (and/or facilities). Clearly, in an investiga-tion of health outcomes as a consequence of physicalrestraint initiation, resident risk factors for physicalrestraint initiation and physical/mental health needto be carefully controlled. This is because someresident factors may be associated with both physicalrestraint initiation and physical/mental health out-comes. Several studies have characterized character-istics of nursing home residents that predict physicalrestraint initiation, and several resident (i.e., de-mographic and clinical) characteristics are consistentin their association with restraint initiation. Theseinclude age, gender, race/ethnicity, cognition, ADLstatus, history of falls, psychotropic medication use,mental health, and ambulatory status (Capezutiet al., 1996; Castle & Mor, 1998).

Given prior findings showing a relationship be-tween the structural characteristics facility size,staffing, staff-to-patient ratios, occupancy rate,ownership, chain membership, and Medicaid censusin the use of physical restraints, we included thesefactors in this analysis as facility-level controlvariables (Davis, 1991).

We also included several market and policyfactors. These variables were significant in studiesthat examined the utilization of other nursing homeservices. They include competition (measured usingthe Herfindahl index) and managed care penetration(Banaszak-Holl & Hines, 1996).

Statistical Analysis

We first analyzed factors that were associatedwith the initiation of physical restraints. Weexamined the association of resident, facility,market, and policy factors with restraint initiationfor a sample of new residents who were unrestrainedfor the first two quarters of residency. By condition-ing on these health status variables measured atadmission and in the second quarter, we accountedfor the residents’ health status and health trend ina very comprehensive way. Next we analyzed theassociation of restraint initiation with subsequentphysical and mental health, controlling for priorphysical and mental health, facility, and marketcharacteristics. We motivated these analyses with


by guest on Decem


Dow

nloaded from


simple empirical models of the decision by a facilityto restrain a resident and of the subsequent change inhealth status.

In a simple model of the decision by a facility torestrain a resident, this is the result of the facility’sorientation (this may be an explicit policy, orculture, or normal operating procedure) to the parti-cular situation of the resident. From the perspectiveof the facility, residents can be characterized by thedegree to which their care would be made easier ifthey were restrained. We call this characteristic R*i,where the subscript i represents an individualresident. Residents will have a higher value of R*iif they are more likely to disrupt the facility andharm themselves or others if left untended andunrestrained. In the empirical work, we represent R*ifor each resident as a linear combination of healthstatus measures, demographic measures, and anunobserved residual.

Each facility has an orientation that it applies toits residents to determine whether a resident shouldbe restrained. The orientation for facility j isrepresented by a cutoff value P*j. A facility willhave a higher value of P*j if it has a more aggressiverestraint orientation. Facility j will restrain all of itsresidents for whom the advantage to the facility ofrestraints exceeds this cutoff: R*i . P*j. Facilitieswill differ in the value they place on restraint use andtherefore have different values of P*j. Formally, themodel can be written as follows:

Ri ¼ 1 if Ri�.Pj� ¼ 0; ð1Þotherwise

Ri� ¼ Xibþ ei; ð2ÞPj� ¼ Zjcþ tj; ð3Þ

where Xi is a vector of measured resident character-istics, Zj is a vector of measured facility character-istics, ei and tj represent unmeasured resident andfacility characteristics, and b and c are coefficientvectors to be estimated. Equation 1 represents theapplication of facility j’s restraint orientation toresident i, resulting in the resident being restrained(Ri = 1) or not (Ri = 0).

The appropriate method for estimating thecoefficients associated with the restraint initiationdecision depends on the assumptions that are maderegarding the distributions of the unmeasuredresident and facility characteristics ei and tj. Theunmeasured resident characteristic ei is modeled witha logistic distribution that is independent of Xi andZj. We model the distribution of tj with a normaldistribution that is independent of the othercharacteristics (Xi, Zj, and ei). This model is referredto as a random effects logit (Robins, 2003). Thismodel accounts for the nesting of residents withinfacilities and provides estimates of the impact ofresident and facility characteristics on the probabil-ity of a resident being restrained.

For the second analysis, we modeled the impact ofrestraints on resident health status with the follow-ing equation:

Hi ¼ Ria1 þXia2 þ Zja3 þ xj þ gi: ð4ÞHere, Hi is any one of various measures of healthstatus measured at a time following the decision ofwhether to restrain resident i. Parameter a1 capturesthe impact of the restraint decision. Equation 4indicates that Hi depends on the same measuredresident and facility characteristics that affect therestraint decision. In particular, the collection ofresident characteristics contained in Xi includes thesame information on health status measured by Hi

although measured prior to the decision of whetherto restrain the resident.

The measures of health status that we used as thedependent variables in Equation 4 were from thefourth quarterly record. Again, we let a quarter pass,rather than use health status from the third quarter,because we wanted to be certain that we took ourmeasures of health status following the imposition ofrestraints for those residents who were restrained.

Results

Table 2 contains the estimates of the logitspecification for physical restraint initiation usingthe Pennsylvania sample. We found that a personalhistory of falls in the first two quarters wasassociated with restraint initiation in the thirdquarter and that an increase in falls from theadmission quarter to the second quarter was furtherassociated with restraint initiation. Likewise, lowADL performance in the first two quarters wasassociated with restraint initiation, and the worsen-ing of ADL performance between the admissionquarter and the second quarter was further associ-ated with restraint initiation. Low cognitive perfor-mance and the absence of pressure ulcers during thefirst two quarters were associated with restraintinitiation, but changes in these measures between theadmission quarter and the second quarter did nothave a significant association with restraint initia-tion. The average use of psychoactive medications inthe first two quarters of residency was associatedwith restraint initiation in the third quarter. De-mographics (age, race, gender) were all insignificant.For-profit facilities, facilities with a low ratio ofnurse aides to nurses, and facilities in counties withhigh managed care penetration were more likely touse physical restraints.

Table 3 shows the eight outcomes examinedsubsequent to physical restraint initiation. We foundthat lower cognitive performance (p , .001), lowerADL performance (p , .001), and more walkingdependence (p , .001) were associated with priorrestraint use. We did not find a significant rela-tionship between physical restraint initiation and

Vol. 48, No. 4, 2008 447

by guest on Decem


Dow

nloaded from


subsequent levels of depression, behavior issues,falls, pressure ulcers, or contractures. Each of theregressions controlled for the level and change inhealth status during the first two quarters ofresidency as well as other resident, facility, andmarket characteristics that potentially had an impacton fourth-quarter health outcomes.

Discussion

In 1989, 44% of nursing home residents werephysically restrained (Rader, 1991). More recentresearch suggests that 11% of nursing home residentsare physically restrained (Sullivan-Marx, Strumpf,Evans, Baumgarten, & Maislin, 1999), and the 2006Nursing Home Compare Web site shows that,

Table 2. Factors That Affect Initiation of Physical Restraint Use

Physical Restraint Use in Third Quarter of Stay

Variable Coefficient SE

Health status variables

CPS (average of first two quarters) 0.269*** 0.050CPS (change) 0.048 0.065ADL scale (average of first two quarters) 0.402*** 0.091ADL scale (change) 0.290*** 0.081Depression (average of first two quarters) 0.009 0.053Depression (change) �0.017 0.042Behavior issues (average of first two quarters) 0.107 0.080Behavior issues (change) �0.005 0.061Falls (average of first two quarters) 1.197*** 0.177Falls (change) 0.202** 0.099Pressure ulcers (average of first two quarters) �0.139 0.231Pressure ulcers (change) �0.308** 0.151Contractures (average of first two quarters) �0.021 0.565Contractures (change) 0.579 0.540Walking dependence (average of first two quarters) �0.302 0.310Walking dependence (change) 0.046 0.234

Resident control variables

Age 0.170 0.163Age squared �0.001 0.001Male 0.097 0.147Black 0.022 0.260Minority (other than Black) �0.337 0.748Psychoactive meds (average of first two quarters) 0.691*** 0.197Psychoactive meds (change) 0.112 0.178

Facility control variables

For-profit ownership 0.426** 0.196RNs per bed �4.854 3.675LPNs per bed �2.269 2.846NAs per bed 1.077 1.457Bed size 0.001 0.001Chain �0.001 0.183Medicaid occupancy 0.206 0.455Occupancy rate �0.295 1.024NAs per nurse �0.278* 0.167

Market control variables

Competition 0.025 0.071County occupancy rate 2.228 1.996Nonprofit market share �0.398 0.522Managed care 1.844*** 0.591Constant �15.400** 6.934

Observations 12,686Number of facilities 637Model degrees of freedom 36.000Log likelihood �1,232.975

Notes: CPS = Cognitive Performance Scale; ADL = activity of daily living; RN = registered nurse; LPN = licensed practicalnurse; NA= nurse aide.

*p , .10; **p , .05; ***p , .01.


by guest on Decem


Dow

nloaded from


Table

3.

Reg

ress

ions

of

Fourt

h-Q

uart

erH

ealt

hO

utc

om

eson

Thir

d-Q

uart

erR

estr

ain

tU

seand

Res

iden

t,Faci

lity

,and

Mark

etFact

ors

Variable

CPS

ADLScale

Depression

BehaviorIssues

Falls

Pressure

Ulcers

Contractures

Walking

Dependence

Restrained

(thirdquarterofstay)

0.381***

(0.054)0.447***

(0.054)

�0.102(0.080)

0.024(0.043)

0.028(0.021)

0.006(0.015)

0.004(0.006)0.102***

(0.019)

CPS(prioraverage)

0.835***

(0.006)0.071***

(0.006)

0.006(0.009)

0.044***

(0.005)

0.016***

(0.002)�0.004***

(0.002)

0.001(0.001)�0.003(0.002)

CPS(chan

ge)

0.237***

(0.009)0.031***

(0.009)

�0.010(0.013)

�0.003(0.007)

0.003(0.003)

0.002(0.002)

�0.000(0.001)

0.000(0.003)

ADLscale(prioraverage)

0.093***

(0.011)0.824***

(0.011)�0.049***

(0.016)

�0.012(0.009)�0.040***

(0.004)

0.021***

(0.003)

0.000(0.001)0.094***

(0.004)

ADLscale(chan

ge)

0.073***

(0.010)0.312***

(0.010)

0.015(0.015)

0.013(0.008)

�0.003(0.004)

0.009***

(0.003)

0.001(0.001)0.042***

(0.004)

Depression(prioraverage)

�0.010(0.007)

0.016**(0.007)

0.577***

(0.010)

0.039***

(0.005)

0.006**(0.003)

�0.000(0.002)

0.001(0.001)0.008***

(0.002)

Depression(chan

ge)

0.005(0.005)

0.006(0.005)

0.132***

(0.008)

0.019***

(0.004)

0.003(0.002)

0.002(0.002)

�0.000(0.001)�0.001(0.002)

Behaviorissues

(prioraverage)

0.084***

(0.012)0.063***

(0.012)

0.072***

(0.018)

0.494***

(0.010)

0.018***

(0.005)

�0.000(0.003)

0.003**(0.001)�0.001(0.004)

Behaviorissues

(chan

ge)

0.019**(0.009)0.025***

(0.009)

0.025*

(0.014)

0.104***

(0.007)

0.007**(0.004)

0.004(0.003)

0.003**(0.001)

0.005*

(0.003)

Falls(prioraverage)

0.043*

(0.025)

0.028(0.025)

0.035(0.037)

0.027(0.020)

0.186***

(0.010)

�0.010(0.007)

0.000(0.003)0.035***

(0.009)

Falls(chan

ge)

0.065***

(0.014)0.078***

(0.014)

0.033(0.021)

0.008(0.011)

0.048***

(0.006)

0.002(0.004)

0.002(0.002)0.015***

(0.005)

Psychoactive

meds(prioraverage)

0.136***

(0.031)

0.058*

(0.031)

�0.013(0.047)

0.075***

(0.025)

0.021*

(0.012)

0.004(0.009)

�0.002(0.004)

0.003(0.011)

Psychoactive

meds(chan

ge)

0.023(0.029)

0.034(0.029)

0.009(0.043)�0.062***

(0.023)

�0.015(0.012)

0.016*

(0.008)

�0.001(0.003)

0.006(0.010)

Pressure

ulcers(prioraverage)

�0.097***

(0.031)�0.008(0.031)

0.050(0.046)

�0.041(0.025)�0.037***

(0.012)

0.268***

(0.009)

�0.001(0.004)�0.007(0.011)

Pressure

ulcers(chan

ge)

0.014(0.021)

0.028(0.021)

�0.017(0.031)

�0.009(0.017)

0.007(0.008)

0.082***

(0.006)

�0.002(0.002)

0.005(0.007)

Contractures(prioraverage)

0.025(0.073)

0.042(0.073)

�0.069(0.109)

�0.018(0.059)

0.002(0.029)

�0.016(0.021)

0.708***

(0.009)�0.041(0.026)

Contractures(chan

ge)

�0.006(0.076)�0.040(0.076)

0.124(0.114)

0.136**(0.062)

�0.006(0.030)

0.060***

(0.021)

0.168***

(0.009)�0.017(0.027)

Walkingdependence

(prioraverage)�0.196***

(0.035)

0.011(0.035)

0.108**(0.052)

�0.016(0.028)

0.064***

(0.014)�0.016*

(0.010)

�0.002(0.004)0.472***

(0.012)

Walkingdependence

(chan

ge)

�0.040(0.026)�0.041(0.026)�0.069*

(0.039)

0.034(0.021)

�0.001(0.010)

0.006(0.007)

�0.003(0.003)0.124***

(0.009)

Age

0.047**(0.019)

0.009(0.019)

0.022(0.028)

�0.001(0.015)

0.005(0.008)

0.007(0.005)

�0.001(0.002)

0.005(0.007)

Age

squared

�0.000**(0.000)�0.000(0.000)

�0.000(0.000)

�0.000(0.000)

�0.000(0.000)

�0.000(0.000)

0.000(0.000)�0.000(0.000)

Male

0.018(0.019)

0.013(0.019)

�0.042(0.029)

0.083***

(0.016)

0.031***

(0.008)

0.017***

(0.005)

0.003(0.002)�0.003(0.007)

Black

0.009(0.032)�0.010(0.032)�0.110**(0.048)�0.061**(0.026)

�0.012(0.013)

0.004(0.009)�0.008**(0.004)

0.004(0.011)

Minority

(other

than

Black)

�0.065(0.082)�0.075(0.082)

�0.027(0.122)

�0.053(0.066)

0.033(0.032)

�0.009(0.023)

�0.006(0.010)�0.027(0.029)

For-profitownership

0.027(0.020)�0.010(0.020)

�0.041(0.030)

�0.018(0.016)

0.001(0.008)

�0.005(0.006)

0.000(0.002)

0.004(0.007)

RNsper

bed

0.228(0.342)�0.291(0.342)

0.691(0.510)

0.046(0.276)�0.428***

(0.136)

0.030(0.096)

�0.030(0.040)�0.099(0.120)

LPNsper

bed

0.048(0.240)�0.177(0.241)

1.436***

(0.359)

0.100(0.194)

�0.115(0.096)

0.042(0.068)

�0.022(0.028)�0.107(0.084)

NAsper

bed

�0.128(0.121)

0.038(0.122)

�0.180(0.181)

�0.147(0.098)

0.087*

(0.048)

�0.053(0.034)

0.001(0.014)

0.011(0.043)

Bed

size

0.000***

(0.000)�0.000(0.000)

0.000***

(0.000)

0.000(0.000)

�0.000(0.000)

�0.000(0.000)

�0.000(0.000)�0.000(0.000)

Chain

0.027(0.018)�0.008(0.018)

�0.035(0.027)�0.033**(0.015)

0.001(0.007)

0.001(0.005)

0.001(0.002)�0.006(0.006)

Medicaidoccupan

cy�0.128***

(0.047)�0.029(0.047)

0.100(0.070)

0.084**(0.038)�0.067***

(0.019)

�0.004(0.013)

0.006(0.006)�0.016(0.016)

Occupan

cyrate

�0.118(0.108)�0.015(0.108)

0.305*

(0.161)

0.103(0.087)

�0.007(0.043)

�0.025(0.030)

0.017(0.013)

0.056(0.038)

NAsper

nurse

0.015(0.013)�0.003(0.013)

0.061***

(0.019)

0.004(0.010)�0.010**(0.005)

0.001(0.004)

0.002(0.001)�0.008*

(0.004)

Competition

�0.003(0.007)

0.005(0.007)

�0.002(0.011)

�0.004(0.006)

�0.004(0.003)

0.000(0.002)

�0.001(0.001)

0.002(0.003)

County

occupan

cyrate

�0.095(0.193)�0.211(0.194)�1.021***

(0.289)

0.017(0.156)

�0.024(0.077)

0.093*

(0.055)�0.096***

(0.023)

0.007(0.068)

Nonprofitmarket

share

0.033(0.054)�0.064(0.054)

�0.094(0.080)

0.005(0.043)

0.012(0.021)

0.012(0.015)

0.000(0.006)�0.004(0.019)

Man

aged

care

�0.027(0.060)�0.064(0.060)

�0.115(0.089)

�0.061(0.048)�0.044*

(0.024)

�0.004(0.017)

0.015**(0.007)�0.021(0.021)

Constan

t�1.642**(0.801)

0.225(0.802)

0.002(1.195)

0.087(0.648)

�0.021(0.318)

�0.307(0.226)

0.124(0.094)�0.152(0.280)

Observations

12,686

12,686

12,685

12,685

12,685

12,686

12,647

12,686

R2

0.71

0.64

0.29

0.28

0.06

0.10

0.38

0.49

Note

s:Stan

darderrors

inparentheses.CPS=

Cogn

itivePerform

ance

Scale;

ADL=

activity

ofdaily

living;

RN

=registered

nurse;

LPN

=licensedpractical

nurse;

NA=

nurse

aide. *p

,.10;

**p

,.05;

***p

,.01.

Vol. 48, No. 4, 2008 449

by guest on Decem


Dow

nloaded from


nationally, approximately 9% of residents are re-strained. However, these figures may underestimatetrue restraint initiation, as counts are often madeduring the day. Restraint initiation may be muchhigher at night, when staffing levels are lower. Evenat 9% to 11%, the potential adverse consequences ofrestraint use create concern that physical restraintsare still overused. Physical restraint use is loweramong nursing homes in many other developedcountries, ranging from 2% to 9% of facilityresidents (Ljunggren, Phillips, & Sgadari, 1997).This, and the fact that restraint use rates in U.S.nursing homes vary widely, suggests that manyfacilities here in the United States could functionwith lower rates of restraint use than are currentlyemployed.

Most significantly, cited potential adverse con-sequences include falls, agitation of residents, andcognitive decline. However, the many perniciouseffects of restraint use are not well investigated. Todate, the health consequences of restraining residentsare unclear. Thus, the results of this study areimportant. We show that adverse consequences ofrestraint initiation exist and are substantial. Weshow that lower cognitive performance, lower ADLperformance, and more walking dependence resultwithin 3 months of restraint initiation. Moreover,these results would appear to have practical as wellas statistical significance. For example, when a resi-dent is restrained, one can expect 5% lower ADLperformance, 10% more walking dependence, and4% lower cognitive performance compared to norestraint use.

We must clearly interpret the practical significanceof our findings based on the residents we examined.That is, these results apply to newly admittedresidents who are not restrained for 6 months.Also, we do not expect the same level of ADL,walking, and cognitive performance decline for eachrestraint episode (i.e., a uniform dose–response). Wealso note that we only found significant findings forthree of the eight outcome measures examined.Nonetheless, the higher expected resident healthoutcomes associated with simply not using physicalrestraints are large.

These findings have practice and policy implica-tions. For nursing homes, decline in resident healthstatus could necessitate the dedication of additionalresources to prevent further decline. Thus, wespeculate that restraint initiation may also necessi-tate subsequent use of additional staff (as a result ofresident decline). From a cost–benefit standpoint, itmay then be more expensive to restrain a residentcompared with not restraining the resident in thefirst place. For policy makers, our findings providefurther justification for implementing restraint re-duction policies and for providing resources topromote restraint reduction. These kinds of activitieswould include additional training, practice guides,and emphasis by state surveyors.

Limitations and Suggestions forFurther Research

Our study combined trunk, limb, and chairrestraint use into a single measure of restraint use.Further research may find that different categories ofrestraints are associated with different residentoutcomes. The possibility that such relationshipsexist necessitates that researchers examine each typeof restraint separately. Even with all MDS data from1 year from Pennsylvania, our sample size lacked thepower for us to perform this analysis.

Our measure of restraint ‘‘use’’ may also havebeen limited. The MDS records use of the fivecategories of restraints as 0=not used, 1=used lessthan daily, and 2= used daily. We combined ‘‘usedless than daily’’ and ‘‘used daily’’ for analysis. Thiswas not a fine-grained measure of intensity of use,but again we were limited by the available data.

Recently, Schnelle and associates (2004) ques-tioned the accuracy of the restraint use MDS items.These authors’ research is based on a very shortperiod of observation, a small sample of residents,and an even smaller sample of facilities, and itcertainly needs further verification. Nevertheless, thisissue needs careful consideration for an analysis suchas ours. Schnelle and associates (2004) surfaced theissue that restraints may be undercounted—but notthat they are either overcounted or miscounted.Thus, for this research we can be fairly certain that ifthe MDS recorded a resident as restrained, then thatresident was in fact highly likely to have beenrestrained. Thus, no confound based on misclassifi-cation is likely with any associations with healthoutcomes for these restrained residents. In addition,any residents who were restrained, but who were notrecorded as such in the MDS, would have beenincluded in our pool of comparison residents. So anybias that exists would likely have served to make ourresults more conservative.

Some authors have also noted that ascertainmentbias may exist with the MDS (Mor, 2004). Simplyput, better facilities may be better at completing theMDS. This bias is likely minimal for restraint use butmore likely for the outcomes we examined. Again,this would likely have served to make our resultsmore conservative.

We were careful throughout our study not toclaim a causal interpretation for our findings.However, in some branches of applied science, theuse of temporal ordering, after matching on priorcharacteristics, is sufficient to make claims ofcausality (Granger, 1969). In the present study, weimplicitly compared two sets of individuals withsimilar existing health conditions, one of whom wasthen placed in restraints and one of whom was not.A causal interpretation would hold that if therestrained group experienced a subsequent declinein health status, then the initiation of restraintscaused this decline.


by guest on Decem


Dow

nloaded from


Such a causal interpretation would not be valid ifrestrained residents were atypical in the progressionof their health conditions. Although we controlledfor two quarters of health outcomes, the possibilityremains that changes in health status during the thirdquarter brought about restraint initiation rather thanarose from them. It is also possible that facilities thatwere more aggressive in their restraint initiation alsodiffered in other dimensions of care that affectedhealth outcomes. If either of these factors arecorrelated with a decline in health outcomes, thenrestraint initiation is said to be endogenous and theestimated effect of restraint initiation on healthoutcomes will not be valid, unless methods are usedthat allow for these correlations.

The most common method for addressing thisproblem in a nonexperimental setting is to use one ormore instrumental variables to identify the impact ofan endogenous treatment. Instrumental variables aremeasured characteristics that have a direct effect onthe treatment (i.e., restraint initiation) but only affectthe outcome through the treatment. Such variableswould induce a variation in restraint initiation that is‘‘random,’’ that is, uncorrelated with the unobservedcomponents in the health outcome equation, andthat can be used to estimate the impact of restraintinitiation on health after controlling for the mea-sured variables that are included in the healthoutcome equation. We propose that this approachbe used in future research by using policy differencesamong states as instrumental variables for restraintuse. Some states, for example, provide training forrestraint reduction, and others report restraint lev-els on long-term care facility report cards. This ad-vanced nonexperimental statistical method wouldallow for causal inferences in a setting such as this inwhich a randomized controlled clinical trial is notfeasible.

Another extension of the current work would beto develop a full dynamic model of health status andrestraint use. Such a model could use all availableinformation from the MDS on changes over time inhealth status for each resident as well as informationabout the initiation and cessation of restraint use.Researchers could use event history methods,especially methods for analyzing repeat events, orother panel data methods to study the dynamics ofrestraint use and discrete changes in health status.However, such methods require careful considera-tion of all events that are determined simultaneouslyand the careful modeling of asymmetries between theassociations with event initiation and with eventcessation. For example, standard methods for theanalysis of time-series cross-section data on restraintuse and health status would imply that the asso-ciation of restraint initiation with health status isequal in magnitude but opposite in sign to theassociation of restraint cessation with health status.We find this implication implausible, and we leavethis complex analysis to future work.

Conclusion

We examined eight mental and physical outcomes3 months post physical restraint initiation. Even aftercontrolling for prior health status and resident,facility, and market factors, we found that restrainedresidents are significantly more likely to exhibit lowcognitive performance, low ADL performance, andmore walking dependence than similar residents whoare not restrained. The magnitude of the findingswould suggest that the benefits to residents of notusing restraints are substantial.

References

Banaszak-Holl, J., & Hines, M. (1996). Factors associated with nursing homestaff turnover. The Gerontologist, 36, 512–517.

Burrows, A. B., Morris, J. N., Simon, S. E., Hirdes, J. P., & Phillips, C.(2000). Development of a Minimum Data Set-based depression ratingscale for use in nursing homes. Age Ageing, 29(2), 165–172.

Burton, L. C., German, P. S., Rovner, B. W., & Brant, L. J. (1992). Physicalrestraint use and cognitive decline among nursing home residents.Journal of the American Geriatrics Society, 40, 811–816.

Capezuti, E., Evans, L., Strumpf, N., & Maislin, G. (1996). Physical restraintuse and falls in nursing home residents. Journal of the AmericanGeriatrics Society, 44, 627–633.

Castle, N. G. (2006). Mental health outcomes and physical restraint use.Administration and Policy in Mental Health, 1–9.

Castle, N. G., & Mor, V. (1998). The use of physical restraints in nursinghomes: A review of the literature since the Nursing Home Reform Act of1987. Medical Care Research and Review, 55(2), 139–170.

Davis, M. A. (1991). On nursing home quality: A review and analysis.Medical Care Review, 48(2), 129–166.

Ejaz, F. K., Folmar, S. J., Kaufmann, M., Rose, M. S., & Goldman, B. (1994).Restraint reduction: Can it be achieved? The Gerontologist, 34,694–699.

Evans, L. K., & Strumpf, N. E. (1989). Tying down the elderly: A review ofthe literature on physical restraint. Journal of the American GeriatricsSociety, 37, 65–74.

Folmar, S., & Wilson, H. (1989). Social behavior and physical restraints. TheGerontologist, 29, 650–653.

Granger, C. (1969). Investigating causal relations by econometric models andcross-spectral methods. Econometrica, 37, 424–438.

Gruber-Baldini, A. L., Zimmerman, S. I., Mortimore, E., & Magaziner, J.(2000). The validity of the Minimum Data Set in measuring the cognitiveimpairment of persons admitted to nursing homes. Journal of theAmerican Geriatrics Society, 48, 1601–1606.

Harrington, C., Zimmerman, D., Karon, S. L., Robinson, J., & Beutel, P.(2000). Nursing home staffing and its relationship to deficiencies. Journalof Gerontology: Social Sciences, 55B, S278–S287.

Hawes, C., Morris, J. N., Phillips, C. D., Mor, V., Fries, B. E., &Nonemaker, S. (1995). Reliability estimates for the Minimum DataSet for nursing home resident assessment and care screening (MDS):Results from the final field testing. The Gerontologist, 35,172–178.

Hawes, C., Vladeck, B. C., Morris, J. N., Phillips, C. D., & Fredeking, H.(2003). The RAI and the politics of long-term care: The convergence ofscience. Milbank Memorial Fund. Retrieved November 16, 2007 fromhttp://www.milbank.org/reports/interRAI/020222interRAI.html.

Levine, J. M., Marchello, V., & Totolos, E. (1995). Progress towarda restraint-free environment in a large academic nursing facility. Journalof the American Geriatrics Society, 43, 914–918.

Ljunggren, G., Phillips, C. D., & Sgadari, A. (1997). Comparisons of restraintuse in nursing homes in eight countries. Age and Ageing, 26(Suppl. 2),43–47.

Lofgren, R. P., MacPherson, D. S., Granieri, R., Myllenbeck, S., & Sprafka,J. M. (1989). Mechanical restraints on the medical wards: Are protectivedevices safe? American Journal of Public Health, 79, 735–758.

Marks, W. (1992). Physical restraints in the practice of medicine: Currentconcepts. Archives of Internal Medicine, 152, 2203–2206.

Mion, L. C., Frengley, J. D., Jakoveic, C. A., & Marino, J. A. (1989). Afurther exploration of the use of physical restraints in hospitalizedpatients. Journal of the American Geriatrics Society, 37, 949–956.

Mor, V. (2004). A comprehensive clinical assessment tool to inform policyand practice: Applications of the Minimum Data Set. Medical Care,42(Suppl. 4), 50–59.

Vol. 48, No. 4, 2008 451

by guest on Decem


Dow

nloaded from


Mor, V., Branco, K., Fleishman, J., Hawes, C., Phillips, C. D., Morris, J. N.,et al. (1995). The structure of social engagement among nursing homeresidents. Journal of Gerontology: Psychological Sciences, 50B, P1–P8.

Moseley, C. B. (1997). The impact of restraints on nursing home residentoutcomes. American Journal of Medical Quality, 12(2), 94–102.

Rader, J. D. (1991). Leveling off restraints. Geriatric Nursing, 13, 71–73.Robins, J. M. (2003). General methodological considerations. Journal of

Econometrics, 112, 89–106.Schnelle, J. F., Bates-Jensen, B. M., Levy-Storms, L., Grbic, V., Yoshii, J.,

Cadogan, M., et al. (2004). The Minimum Data Set prevalence ofrestraint quality indicator: Does it reflect differences in care? TheGerontologist, 44, 245–255.

Schnelle, J. F., Simmons, S. F., & Ory, M. G. (1992). Risk factors that predictstaff failure to release nursing home residents from restraints. TheGerontologist, 32, 767–770.

Snowden, M., McCormick, W., Russo, J., Srebnik, D., Comtois, K., Bowen,J., et al. (1999). Validity and responsiveness of the Minimum Data Set.Journal of the American Geriatrics Society, 47, 1000–1004.

State Operations Manual. (2002). Washington, DC: Centers for Medicareand Medicaid Services. Retrieved January 12, 2005 from www.cms.hhs.gov/manuals/pub07pdf/pub07pdf asp.

Straker, J. (1999). Reliability of OSCAR occupancy, census, and staff data: Acomparison with the Ohio Department of Health Annual Survey of Long-Term Care Facilities (Tech. Rep. Series 3-01). Oxford, OH: MiamiUniversity, Scripps Gerontology Center.

Sullivan-Marx, E. M., Strumpf, N. E., Evans, L. K., Baumgarten, M., &Maislin, G. (1999). Predictors of continued physical restraint use innursing home residents following restraint reduction efforts. Journal ofthe American Geriatrics Society, 47, 342–348.

Tinetti, M. E., Wen-Liang, L., Marottoli, R. A., & Ginter, S. F. (1991).Mechanical restraint use among residents in skilled nursing facilities.Journal of the American Medical Association, 265, 468–471.

Werner, P., Cohen-Mansfield, J., Braun, J., & Marx, M. S. (1989). Physicalrestraints and agitation in nursing home residents. Journal of theAmerican Geriatrics Society, 37, 1122–1126.

Received May 22, 2007Accepted September 21, 2007Decision Editor: William J. McAuley, PhD


by guest on Decem


Dow

nloaded from


Physical Restraint Initiation in Nursing Homes and Subsequent Resident Health

Documents

Transcript of Physical Restraint Initiation in Nursing Homes and Subsequent Resident Health