Physical restraint use among nursing home residents: A comparison of two data collection methods

Nursing homes have the mandate to offer care settings to frail dependent older individuals. However, a renewed emphasis has emerged over the past decades to become more than just a home for older people [1]. A growing number of facilities are actually striving to preserve residents' sense of control and dignity in order to achieve the highest level of well-being [2, 3]. This new way of thinking is based upon values of respect of autonomy and freedom for older persons in various ways such as the resident's right to take risks or to make his/her own choices [4].

Although predominantly intended as protective devices, physical restraints in nursing homes are being denunciated as measures that go conversely with the aforementioned principles [5]. Justifications for controlling confusion, agitated and aggressive behaviors are being questioned [6–8] and beneficial effects of physical restraints on falls and injuries, incontinence, muscle atrophy and quality of life challenged [8–15]. Moreover, physical restraints have been associated with cognitive impairment, nosocomial infections, pressure sores and death [10, 16–19].

According to the literature, the overall prevalence of restraint use in nursing homes ranges between 4 and 68% [5, 20]. This wide variation may be explained by definitions of physical restraints used, study sample sizes, characteristics of care settings, and residents' characteristics and cognitive status. Another explanation could be the choice of techniques of data collection. Several methods have been used alone or in combination for the measurement of physical restraint use [20–22]: direct observation, survey or interview of nursing staff, review of medical and nursing records and, when the cognitive status allows it, interview with residents themselves. In view of the consequential issues surrounding the use of physical restraints, it is important to have a method of measurement as valid and reliable as possible. While direct observation is undoubtedly the most valid and reliable method of measurement, it is also the most expensive means to measure physical restraint use. On the other hand, abstracting data from medical records and interviewing nursing staff have the potential to reduce the cost associated with data collection, but their sensitivity and specificity values need to be demonstrated. In addition, apart from the USA, data sources such as Minimum Data Set (MDS) have not been widely implemented in nursing home facilities throughout the world.

The objective of this study was to determine the sensitivity and specificity of the measurement of physical restraint use reported by members of the nursing staff and reviewed from medical and nursing records among nursing home residents, compared to direct observation. Since underreporting is much more susceptible to be problematic than overreporting, another objective of this study was to compare the sensitivity of the information reported by one nurse with that reported by two nurses or more questioned together. Our research hypothesis was that sensitivity of the interview is highest when the information is collected from more than one nurse.

The study was conducted in eight facilities representing a convenience sample of the long-term care facilities in the Quebec City area, Canada. These institutions were carefully selected in order to include a mix of characteristics in size (small and large), geographic location (urban and rural), university affiliation and vocation (units associated with psychiatric or rehabilitation team). Selection was made after discussion with nursing direction of each setting to gather units of different practice such as regular units and specialized units for residents with dementia or severe behavioral problems. Twenty-five subjects were randomly chosen from each unit; if an unit comprised less than 25 residents, all of its residents were included. This study was approved by the ethics committee at Laval University. Data collection took place between January and June 1992.

Data collection was carried out in 16 nursing units. Of these units, eight depicted skilled nursing home care units, seven nursing home care units, and one long-term care unit within a short-term care hospital. Information was collected for 377 residents with the help of 28 nurses. Residents' age ranged from 32 to 102 years, with a median of 80 years. The sample was 62% female, and median length of stay was 45 months (0 to 720 months). Benzodiazepines and neuroleptics were administered to 35% and 25% of the subjects, respectively.

The measurement of physical restraint use according to interview with members of the nursing staff and review of medical charts and nursing orders both reflect accurately the reality observed in long-term care setting residents. Our study has also shown that sensitivity and specificity values of the reported measurement are higher than those calculated from medical charts and nursing orders. This phenomenon is not surprising considering that the keeping of medical and nursing orders in nursing homes isn't usually done on a daily basis [27], as opposed to acute care settings.

The current investigation was carried out in units of diverse facilities. The selection of these facilities was intended to allow the participation of subjects and care units of various characteristics as compared to other studies usually designed [28]. The sample of nursing home residents included in this study corresponded well to the physically and cognitively impaired residents generally housing in long-term care institutions.

Limitations of the current study must be taken into account when interpreting these findings. First, data were collected in 1992. Due to the implementation of the OBRA act, it is probable that the prevalence figures given in the current study are overestimations of those that would be observed in 2004. On the other hand, the province of Quebec just recently launched its first comprehensive policy on physical restraint use [23]. Furthermore, the purpose of this study was to compare the sensitivity values of two reporting techniques with direct observation. This comparison should not be affected by the prevalence of physical restraint use. In addition, although a higher proportion of restrained residents might seem more difficult for the nurses to remember as compared to a lower proportion, the nurses didn't show any hesitation when recalling the use of physical restraints as the majority of residents had been living there for a long period of time. Second, we used a convenience sample of long-term care facilities rather than one drawn randomly. We wanted to determine differences and similarities in various practice facilities regarding physical restrain use. The chosen sample provided a relatively broad range of clinical settings. Also, the assessment by nurses was performed the day after direct observation. This time period was chosen in order to reduce recall bias as much as possible, and therefore increase the sensitivity of the reporting technique although this may not be practical in many situations. Another limitation for the interpretation is the use of a descriptive study design. This design is useful to measure the frequency in which a situation occurs or collect data on possible risk factors, but does not allow to infer causal relationships.

It is well known that the prevalence of residents with physical restraints is usually underreported since a social desirability bias tends to affect the validity of the information when the nursing staff has to declare the use of restraints [29]. Despite that restraints are generally applied for safety reasons, nurses nevertheless experience inner struggle when they have to apply them [11, 22]. These feelings could influence the nurses' answer when they are interviewed individually and could introduce subsequently a differential misclassification error. In our study, although the sensitivity value improved when interviews were done with two nurses instead of one, the difference was not considered clinically significant. Reported use of physical restraints by two nurses reduced but did not eliminate the presence of an information bias, since the underreporting went down from 14.9 to 5.9%. On the other hand, we do not think that the resulting effect on the prevalence estimates is of consequence. This phenomenon is equally present but to a much lesser extent in the over reporting data since from 6.2% with one nurse, the prevalence of over reported restraint use was reduced to 1.1% when two nurses were interviewed.

Even though other studies have observed an association between residents' characteristics and the risk of being restrained [8, 10, 17, 30, 31], these characteristics were not related to the use of physical restraints. Rather, justifications for the use of physical restraints such as disruptive behaviors (e.g. aggressiveness, agitation, and body alignment problems) were associated with their use. For example, it is noteworthy that the risk for falls as perceived by nursing staff was associated with restraint use whereas a history of falls was not. This means that physical restraints were used as preventive devices for a large proportion of subjects, in spite of numerous studies that do not support such practices [13–15, 22].

Compared to review of clinical records, reported physical restraint use by interviewing nursing staff is a simple, efficient, and valid technique of collection of data regarding their use in nursing homes. No severe information bias was observed even though the use of physical restraints may be associated with poor quality of care. This method of measurement appears to be reliable and valid for research purposes. Moreover, our study provides support to the American initiative in regard to the monitoring of several outcomes in nursing homes through nursing staff reports [32–35]. According to our results, interviewing nursing staff is a sensitive and specific method of eliciting information on physical restraint use. Finally, these results have implications for future research in the field. Interviewing nurses on different aspects of medical and nursing care seemed to be a reliable method.