Three-level Random Intercept Model Research Articles

Gender balance in the workforce and abortion attitudes: A cross-national time-series analysis.

This article explores the relationship between gender balance in the workforce and attitudes towards abortion worldwide. Studies on macro-level conditions related to abortion attitudes overlook the role of gender balance in the workforce-specifically the degree of female representation in a country's workforce. There are strong reasons why this factor could shape abortion attitudes. We argue that such a gender balance creates necessary conditions to break with traditional, anti-abortion ideology and facilitates dissemination and public acceptance of pro-choice views. We test this argument with two different datasets - the Integrated Values Survey and three waves of the International Social Survey Programme-along with two outcomes: general tolerance towards abortion and tolerance towards abortion for pregnant women of low income. Using three-level random intercept models and multiple controls for individual and country-level conditions, the results support our hypothesis: In countries with higher gender balance in the workforce, individuals display higher tolerance towards abortion.

Estimation quality and required sample sizes in three-level contextual analysis models

In multilevel analysis, Level-1 predictors that also explain variance at a higher level are called contextual predictors. In the multilevel manifest covariate model, the Level-2 component is modeled as the average of the Level-1 predictor scores within a cluster. In the multilevel latent covariate model, the predictor is decomposed into two latent variables at Level-1 and Level-2. Performance conditions of these modeling approaches for three-level models are largely unexplored. We investigate the two approaches’ performance with respect to bias, coverage, and power in a three-level random intercept model. Results reveal differences in estimation quality and required sample sizes. We provide sampling recommendations for both approaches.

Adverse childhood experiences among early childhood educators and improving teacher health and well-being: Changes in teacher affect

Children's social-emotional and cognitive development is enhanced when early childhood educators display positive affect and respond appropriately to children's needs. However, there is little research on factors that influence teachers’ positive affect in the classroom, such as past adversity exposure and promoting teachers’ well-being. The current study examined the interactive effect of teacher participation in a well-being program (Be Well Care Well; BWCW) and their exposure to adverse childhood experiences (ACEs) on changes in observed teacher affect over time within the classroom setting. Participants included 95 early childhood teachers who completed the ACEs measure, a live classroom observation before participation in activities, and a post-intervention observation. Multilevel modeling estimated the interactive effects of teacher ACEs, participation in BWCW activities, and time on the dependent variable (i.e., teacher affect). Results from the three-level random intercept model showed a significant three-way interaction. Individuals with a higher number of ACEs and lower participation in the well-being program showed a decrease in positive affect over time. This effect was not significant for teachers who reported fewer ACEs. Results indicate that the degree of participation in the well-being program was particularly salient for teachers with higher exposure to past ACEs. Findings suggest that fostering teachers’ physical health and well-being can enhance their affect and enthusiasm in the classroom setting, particularly for those with higher levels of past adversity exposure.

The urban\u2013rural differential in the association between household wealth index and anemia among women in reproductive age in Ethiopia, 2016

BackgroundAnemia is more prevalent among women, and it is a moderate public health problem in Ethiopia. The wealth status and place of residence of a woman have implications on the intervention of anemia. Studies that examined the relationship between women’s wealth index status and residency in Ethiopia are scarce. We aimed to identify the urban–rural differential in the association between household wealth index and anemia among women of childbearing age in Ethiopia.MethodA cross-sectional design was employed with a nationally representative sample of 14,100 women aged 15–49-year-old from the Ethiopian demographic and health survey conducted in 2016. We used the two-stage sampling method to select the sample size. The primary outcome was anemia in women of childbearing age. A hemoglobin level of below 11 g/dl for pregnant women and 12 g/dl for non-pregnant women was the indicator of anemia. Using a three-level random intercept model to explore associated factors at the individual and household levels quantified the observed and unobserved variations between household wealth index and residence on anemia.ResultsWomen belonging to a lower household wealth index category were more anemic (29.6%) than those middle and above wealth index categories. Women who lived in rural areas (25.5%) were prone to anemia than those who lived in urban areas (17.5%). The odds of anemia were significantly higher in women of the low household wealth category who living in rural compared to women of the middle and above household wealth category who living in urban (AOR = 1.37, 95% CI 1.14–1.65, P < 0.001).ConclusionIn this study, anemia is more common among women who live in rural with the low house wealth category. Therefore, novel public health interventions should target women who live in rural areas with the lowest household wealth status.

A gender-focused multilevel analysis of how country, regional and individual level factors relate to harm from others’ drinking

Background The aim of this study was to examine how gender, age and education, regional prevalence of male and female risky drinking and country-level economic gender equality are associated with harms from other people’s drinking. Methods 24,823 adults in 10 countries were surveyed about harms from drinking by people they know and strangers. Country-level economic gender equality and regional prevalence of risky drinking along with age and gender were entered as independent variables into three-level random intercept models predicting alcohol-related harm. Findings At the individual level, younger respondents were consistently more likely to report harms from others’ drinking, while, for women, higher education was associated with lower risk of harms from known drinkers but higher risk of harms from strangers. Regional rate of men’s risky drinking was associated with known and stranger harm, while regional-level women’s risky drinking was associated with harm from strangers. Gender equality was only associated with harms in models that did not include risky drinking. Conclusions Youth and regional levels of men’s drinking were consistently associated with harm from others attributable to alcohol. Policies that decrease the risky drinking of men would be likely to reduce harms attributable to the drinking of others.

Compositional and Contextual Associations of Social Capital and Fear of Crime

ABSTRACT This study examines the association between individual and district levels of social capital and fear of crime using multilevel analysis. The study uses the 2011 Seoul Survey. Three-level random intercept models were conducted. This study found that only generalized trust in others at the individual level was associated with fear of crime. The current study further compared the variance in fear of crime attributable to the individual, household, and district levels using the intraclass correlation coefficient. The results show that a relatively large proportion of variance in fear of crime can be attributed to the household level, while only a small proportion of variance can be attributed to the district level. The results suggest that it is worth investigating household-level variables in-depth to better explain the fear of crime. Furthermore, the results indicate that social interventions lowering fear of crime are unlikely to be efficient if they target the district level only.

Quality of life improved for patients after starting dialysis but is impaired, initially, for their partners: a multi-centre, longitudinal study

BackgroundQuality of life (QOL) is important to patients with end stage renal disease and their partners. Despite the first 12 weeks being a critical time in the treatment pathway, limited research exists which examines how the transition onto dialysis impacts QOL. In this study we measured QOL in patients and their partners at pre-dialysis and over the first 12 weeks on dialysis to investigate QOL during this crucial period.MethodsPatients and their partners, recruited from 10 renal units in England, completed questionnaires at pre-dialysis (n = 166 participants, 83 couples), 6 weeks (n = 90 participants, 45 couples) and 12 weeks (n = 78, 39 couples) after starting dialysis. On each occasion participants completed a QOL questionnaire (WHOQOL-BREF). Multilevel modelling accommodated the nested structure of couples with repeated measures within participants. Three-level random intercept models estimated changes in WHOQOL general QOL and its four domains (Physical, Psychological, Social and Environment). Two-level random intercept models assessed the relationship between baseline clinical and socio-demographic variables with changes in general QOL.ResultsPatients reported positive changes in general QOL from pre-dialysis to 6 weeks (β = 0.42, p < 0.001, 95% CI 0.19, 0.65) and from pre-dialysis to 12 weeks (β = 0.47, p < 0.001, 95% CI 0.24, 0.71). Partners’ general QOL decreased significantly from pre-dialysis to 6 weeks (β = − 0.24, p = 0.04, 95% CI -0.47, − 0.01) but returned to its original level at 12 weeks. Patients reported improvements in the physical domain between pre-dialysis and 12 weeks (β = 6.56, p < 0.004, 95% CI 2.10, 11.03). No other domains changed significantly in patients or partners. Only in patients were there significant associations between moderator variables and general QOL. High comorbidity risk level and diabetes were associated with poorer QOL at pre-dialysis whereas being female and having an arteriovenous fistula were linked with improvements in general QOL.ConclusionsPatients reported significant improvements in their general and physical QOL after starting dialysis. Partners’ general QOL worsened after patients started dialysis but improved by 12 weeks. Both patients and partners may benefit from additional educational and counselling services in the lead up to, and immediately after starting dialysis, which could facilitate the transition onto dialysis and improve QOL in both.Study registrationThis study was adopted on the NIHR Clinical Research Network (UK). The details of this study are registered on the Research Registry website (www.researchregistry.com). The identifier for this study is researchregistry2574.

The development of a goodness-of-fit test for high level binary multilevel models

Before making inferences about a population using a fitted model, it is necessary to determine whether the fitted model describes the data well. A poorly fitted model may lead to biased and invalid conclusions, resulting in incorrect inferences. Recent studies show the necessity of goodness-of-fit tests for high level binary multilevel models. The focus here was to develop a goodness-of-fit test to use in the model adequacy testing of high level binary multilevel models and to examine, whether the type I error and power hold for the newly developed goodness-of-fit test considering a three-level random intercept model.

A multilevel approach to correlates of anaemia in women in the Democratic Republic of Congo: findings from a nationally representative survey.

Anaemia accounts for a significant proportion of pre- and post-partum morbidity and mortality in low-income countries with sequelae, including an increased risk of infection. Factors contributing to anaemia need to be addressed through the introduction of evidence-based measures to control and prevent the disease. We aimed to determine the prevalence of anaemia in women of child-bearing age in the Democratic Republic of Congo (DRC) and investigate the associated individual, household and community level factors. Cross sectional representative population data from the 2013-2014 DRC Demographic and Health Survey (DHS) was used. The primary outcome was anaemia in women, stratified according to pregnancy in those of child-bearing age. A haemoglobin level of below 11 g/dl for pregnant women and 12 g/dl for non-pregnant women was used as the indicator of anaemia. Using a three-level random intercept model this study explored risk factors at individual, household and community levels and quantified the observed and unobserved variations between households and communities. Thirty-eight percent of women in the DRC are anaemic. Anaemia is significantly higher in younger, pregnant and underweight women, as well as those with comorbidities, including HIV and malaria who are living in the capital city Kinshasa. Anaemia varies within and between households and communities in the DRC. Integrated approaches to reduce anaemia in settings with high malaria and HIV prevalence such as the DRC should target households.

Keel bone differences in laying hens housed in enriched colony cages.

Keel bone damage may be painful to birds and affect their production. In order to better understand the frequency, position, and timepoint of keel bone damage that occur during production, the integrity of W-36 laying hen keel bones housed in enriched colony cages at 748.4 cm2 (116 in2) was evaluated. At four time points, 120 birds (10 per cage; three cages per each of four rooms) had keel bones evaluated. Each hen was placed in a motion limiting restraint, scanned using computed tomography (CT), fitted in vests containing tri-axial accelerometers, and placed back in their cages for 21 d. After 21 d, the hens were rescanned and returned to their cages. This process was repeated after 133 d. The CT scans were imported into Mimics analysis software (Materialise, Plymouth, MI, USA); 3D models were made of each keel bone at each time point and exported to 3-matic analysis software (Materialise, Plymouth, MI, USA). Each laying hen's keel bone model was superimposed onto scans from multiple time points resulting in four bone pairings representative of each 21-d period, the 133-d period, and the entire duration of the project. Next, the proximal portion of each bone pairing was edited to normalize bone shape according to a strict protocol. Additionally, each pairing was divided into three portions: distal aspect (3cm), proximal aspect (2cm), and middle portion (remaining). Whole bone pairing and each bone portion was analyzed using the Part Comparison tool in 3-matic. Raw data were compiled into three datasets and analyzed in SAS 9.3 using the GLIMMIX procedure using a three-level random intercept model. The model controlled for time, part, part(time), and system with random intercepts of bird(cage) and cage. Overall, results revealed that the greatest morphological changes occurred during the first 21-d period with regards to time (P = 0.03) and in the distal aspect of the keel with regards to part (P < 0.0001).

Interviewer variance and its effects on estimates

PurposeIn survey methodology, it is well-known that interviewers can have an impact on the registered answers. This paper aims to focus on one type of interviewer effect that arises from the differences between interviewers in the systematic effects of each interviewer on the answers. In the first case, the authors evaluate interviewer effects on the measurement of alcohol consumption in European countries. The second case is about the interviewer effects on the respondents’ tendency to (non)differentiate their responses and the consequences of this response style for the correlation between variables.Design/methodology/approachThe interviewer effects are evaluated by means of interviewer variance analysis. Because respondents are nested within interviewers, we can specify a two- or three-level random intercept model to calculate the proportion of variance explained by the interviewers. Data from the seventh round of the European Social Survey are used.FindingsThe results in the first case show that the substantive conclusions about the effect of gender and education on the alcohol measures continue to hold when interviewer effects are taken into account. The results of the second case make clear that interviewer effects on attitudinal questions are considerable. There is also a significant effect of the interviewers on the degree that respondents differentiate their responses. The results also illustrate that correlations between attitudinal variables are influenced. This also implies that the results of statistical procedures using a correlation or covariance matrix can be strongly influenced by the tendency to (non)differentiate and the interviewers’ impact on this tendency.Originality/valueThe results clearly demonstrate that there are considerable differences between countries concerning the impact of the interviewers on substantive variables. Cross-national differences are striking and the importance and necessity to evaluate interviewer effects in a cross-national survey becomes clear.

Environmental policy performance and its determinants: Application of a three-level random intercept model

This paper contributes to the study of environmental and energy policy by using a three-level random intercept (TLRI) model to rank the performance of different countries. Inspired by the literature on Item Response Theory and multilevel latent models, the TLRI model treats policy commitment as a latent variable which is estimated conditional on the difficulty of the policy portfolio implemented by each country. This approach is characterized by three novel aspects. First, the model results in a ranking of countries which is conditional on the complexity of their chosen policy portfolio. Second, it provides a unified framework in which to construct a policy indicator and to study its determinants through a latent regression approach. The resulting country ranking can thus be cleaned from the effect of economic and institutional observables which affect policy design and implementation. Third, the model estimates parameters which can be used to describe and compare policy portfolios across countries. We apply this methodology to the case of energy efficiency policies in the industrial sectors of 29 EU countries between 2004 and 2011. In the conclusions we highlight the future possible applications of this approach, which are not confined to the realm of environmental and energy policy.

Effects of post-abortion family planning services on preventing unintended pregnancy and repeat abortion (INPAC): a cluster randomised controlled trial in 30 Chinese provinces

BackgroundAround 10 million induced abortions are conducted annually in China; a third of the women having had those abortions have undergone repeat abortions. Most abortions are performed in hospital settings in which post-abortion family planning (PAFP) services are often lacking. This study aims to evaluate the effects of integrating PAFP services into abortion services on the reduction of unintended pregnancy and repeat abortion in China. MethodsThis was a three-arm cluster (hospital) randomised controlled trial. Study participants were women undergoing an abortion within 12 weeks of pregnancy. 90 hospitals were selected from 30 Chinese provinces and allocated randomly (1:1) into two intervention groups or one control group. Intervention group 1 included provision of family planning information, contraceptive counselling, involvement of the male partner, and free provision of contraception; intervention group 2 included incentive mechanisms for health-care providers in addition to group 1; and the control group received normal care, with no intervention. Eligible women were followed up for 6 months. The primary outcomes were the rates of unintended pregnancies and repeat induced abortions. We used a three-level random intercept model to estimate the effects of intervention using a generalised linear mixed model, and we used SAS PROC GLIMMIX with maximum likelihood with Laplace approximation to perform this multilevel modelling approach. This study received ethical approval from the Ethical Committees at Ghent University, Belgium, on May 26, 2014 (B670201421116), and from the National Research Institute for Family Planning, China, on March 6, 2014. All participants provided a Chinese written informed consent. This trial has been registered at International Standard Randomised Controlled Trial, number ISRCTN01846583. FindingsWe recruited 17 235 eligible women from July 11, 2014, to Aug 20, 2015. The intervention 1 group included 5856 women, intervention 2 group included 5791 women, and the control group included 5588 women. The proportion of patients who were followed up for up to 6 months were similar in the three groups (74·7% [4372 of 5856] in intervention group 1, 77·1% [4466 of 5791] in intervention group 2, and 75·7% [4231 of 5588] in the control group). The proportion of patients with unintended pregnancy within 6 months after abortion was lower in both intervention groups (1·2% [59 of 5011] in group 1 and 1·2% [58 of 4986] in group 2) than in the control group (3·2% [155 of 4817]). The proportion of patients who had a repeat abortion was 0·9% (45 of 5011) for the intervention group 1, 0·8% (41 of 4986) for the intervention group 2, and 1·6% (77 of 4817) for the control group (ie, 16–38% lower in the interventions group than that in the control group). Statistical analysis is ongoing. InterpretationIntegrating post-abortion family planning services into hospital-based abortion services could decrease unintended pregnancy and repeat abortions. A policy and guidelines on integration of PAFP into routine abortion services is urgently needed to reduce the number of abortions. FundingEuropean Commission FP7 (282490).

Household food insecurity and future orientation of Ghanaian youth and their parents

ABSTRACTFood insecurity heightens risk for poor mental health and psychosocial functioning. Higher levels of future orientation influence numerous desirable behaviors. However, limited evidence exists on the association between food insecurity and future orientation, particularly in low-resource settings. The objective of this study was to examine the association between food insecurity and future orientation of Ghanaian youth and their parents. The study included a cross-sectional sample of 2656 youth and 2656 parents from 8 of Ghana’s 10 regions. Food insecurity was measured using the Household Food Insecurity Access Scale. Future orientation in the Ghanaian context was measured using three distinct factors adapted from the School Success Profile and the Consideration of Future Consequences scale. We analyzed our data using hierarchical linear modeling, with a three-level linear random-intercept model with covariates. Results suggest an inverse relationship between food insecurity and future orientation of youth and their parents. Food insecurity was consistently and significantly associated with lower orientation toward success (β = −0.18, 95% confidence interval [CI] = −0.22, −0.14) and higher uncertainty of the future (β = 0.22, 95% CI = 0.17, 0.28) among youth, as well as lower consideration of future consequences (β = −0.15, 95% CI = −0.23, −0.06) among parents. Additionally, severe food insecurity was associated with the lowest future orientation scores for youth and their parents. Our findings indicate that, under conditions of food insecurity, youth and their parents lose sight of the future, which may come at a great cost. Encouraging individuals to engage in thinking about the future or to create future images of their selves may not be meaningful if their basic needs, including access to food, are not met. Programs that provide opportunities to generate income and accumulate assets may have a twofold effect of increasing access to food and improving future orientation.

The Effects of Unemployment and Insecure Jobs on Well-Being and Health: The Moderating Role of Labor Market Policies

Labor market insecurities have been growing in Europe and previous research has illustrated that unemployment and insecure jobs negatively affect individuals’ well-being and health. Although empirical evidence suggests that these effects vary substantially across different welfare states, we still know little about the moderating role of specific labor market policies. Taking a cross-national comparative perspective, this article investigates how passive and active labor market policies (PLMP, ALMP) as well as employment protection legislation (EPL) shape the experience of unemployment and insecure jobs. We complement micro data of round 1–6 (2002–2012) of the European Social Survey with time-varying macro indicators of PLMP, ALMP, and EPL. The data include about 89,000 individuals nested in 112 country-rounds and 26 countries respectively. We apply three-level random intercept models as well as pooled linear regression models including country fixed effects. The results show that labor market policies are important in shaping the experience of unemployment, but are less relevant for workers in insecure jobs. Specifically, higher unemployment benefit generosity buffers the negative effects of unemployment on well-being but not health. Moreover, we discuss different interpretations for the finding that higher ALMP expenditures are associated with more negative effects of unemployment on well-being and health. With respect to EPL it is found that in countries with high insider protection, deregulating the restrictions on the use of temporary employment increases the negative effects of unemployment on well-being and health.

Interviewer Effects on Non-Differentiation and Straightlining in the European Social Survey

Abstract In this article we examine the interviewer effects on different aspects of response styles, namely non-differentiation and straightlining, which in general refers to the tendency to provide the same answers to questions in a block of questions. According to research about response styles, the impact of the interviewer on this kind of response behavior is rare. Five blocks of items in the questionnaire in the sixth round of the European Social Survey (2012) are used in the analysis. These data also allow for an evaluation of the differences between countries in terms of non-differentiation and straightlining. Five different measurements of these aspects of response style are used in the analysis. To disentangle the impact of respondents and interviewers on these aspects of response style, a three-level random intercept model is specified. The results clearly show interviewer effects on the respondent’s tendency to select a response category that is the same as the response category for the previous item. In some countries the proportion of explained variance due to differences between interviewers is larger than the proportion of variance explained by the differences between respondents.

Generalized SAMPLE SIZE Determination Formulas for Investigating Contextual Effects by a Three-Level Random Intercept Model.

Behavioral and psychological researchers have shown strong interests in investigating contextual effects (i.e., the influences of combinations of individual- and group-level predictors on individual-level outcomes). The present research provides generalized formulas for determining the sample size needed in investigating contextual effects according to the desired level of statistical power as well as width of confidence interval. These formulas are derived within a three-level random intercept model that includes one predictor/contextual variable at each level to simultaneously cover various kinds of contextual effects that researchers can show interest. The relative influences of indices included in the formulas on the standard errors of contextual effects estimates are investigated with the aim of further simplifying sample size determination procedures. In addition, simulation studies are performed to investigate finite sample behavior of calculated statistical power, showing that estimated sample sizes based on derived formulas can be both positively and negatively biased due to complex effects of unreliability of contextual variables, multicollinearity, and violation of assumption regarding the known variances. Thus, it is advisable to compare estimated sample sizes under various specifications of indices and to evaluate its potential bias, as illustrated in the example.

Environmental Policy Performance and its Determinants: Application of a Three-Level Random Intercept Model

We propose the use of a three-level random intercept model to measure the degree of environmental policy performance of different countries and to study its determinants. Inspired by the literature on multilevel latent models and Item Response Theory (IRT), this framework treats policy commitment as a latent variable which is estimated conditional on the difficulty of the policy portfolio implemented by each country. We contribute to the study and scoring of environmental and energy policies in three main ways. First, the model results in a ranking of countries which is conditional on the complexity of their chosen policy portfolio. Second, we provide a unified framework in which to construct a policy indicator and to study its determinants through a latent regression approach. The resulting country ranking can thus be cleaned from the effect of economic and institutional observables which affect policy design and implementation. Third, the model estimates parameters which can be used to describe and compare policy portfolios across countries. We apply this methodology to the case of energy efficiency policies in the industrial sectors of 29 EU countries between 2004 and 2011. We conclude by highlighting the future possible applications of this approach, which are not confined to the realm of environmental and energy policy.

Demand for voluntary basic medical insurance in urban China: panel evidence from the Urban Resident Basic Medical Insurance scheme

This paper investigates the key factors associated with the demand for Urban Resident Basic Medical Insurance (URBMI), which was established in 2007 and aims to cover all Chinese urban residents. Two waves of longitudinal household survey data are used, and a three-level random-intercept logit model is used for the analysis. Two different sets of explanatory variables were identified for adults and children, separately. Results suggest for both the adult and the child samples that income, health status, age and health risk behaviours are key influencing factors for basic medical insurance demand. The household head's characteristics are also significantly related to other household members' medical insurance demands. Specifically, household heads who are more educated or retired are more likely to purchase medical insurance for their children. These findings suggest that an expansion of the special subsidy to the poor or, probably more important, a risk-adjusted benefit package may be needed for voluntary basic medical insurance in China. In addition, adverse selection consistently exists and is a major challenge for the sustainability of medical insurance financing. To expand insurance coverage for children, especially those under school age, special efforts (possibly through health education or health promotion) should be focused on the household head, particularly those engaging in risky health behaviours.

Use of physical restraints in nursing homes and hospitals and related factors: a cross‐sectional study

The aim of the study was to investigate factors related to the use of restraints and to explore whether the rate of nurses was an influencing factor regarding the use of restraints in German nursing homes and hospitals. Restraints are frequently used measures in hospitals and nursing homes. Risks for falls and small nurse workforces are discussed in relation to the use of restraints. A secondary analysis of a cross-sectional study was carried out. Methods. Data were collected by trained nurses using standardised questionnaires in 76 nursing homes (n = 5521) and 15 hospitals (n = 2827). For data analysis, a 3-level random intercept logistic model was used. The prevalence of restraints (bed rails and/or belts) was 9·3% for hospital patients and 26·3% for nursing home residents. Amongst hospital patients, restraint use was more prevalent in women, older patients, patients with a high care dependency, patients who fell during the last two weeks, patients with a perceived risk of falls, polypharmacy, urinary incontinence, disorientation and confinement to bed. In the nursing homes, the restrained residents were significantly younger, more care dependent, had less falls and were more often urinary incontinent, disoriented and bedfast. The rate of nurses was not significantly related to the use of restraints in hospitals, and nursing homes according to the three-level random intercept model. Hospital patients with previous falls were more often restrained, but in the nursing homes, the restrained residents experienced less falls. The number of qualified nursing staff had no significant influence on the use of physical restraints. Lower nurse staffing ratios were not related to higher frequencies of restraint use in this study.