Lynsie Ranker

12/13/12

SafeHip Soft Evaluation Proposal

Study Participants:

Target Population: Long-stay nursing home residents age 70 and older (further criteria and justifications below)

Inclusion and Exclusion Criteria (and reasoning):

  • Must be a nursing home resident (studies suggest that elderly who are in nursing home facilities are at greater risk of hip fracture than elderly of the same age and sex who live independently.1,2)
  • Must be age 70 or older at time of study enrollment.3 (risk of falls is higher in this population than among 65 and older3)
  • Must have lived in the nursing home for three months or more3 (to assure they are likely to be long-term stay residents)
  • Must have at least attempted to walk without assistance in the past month (allows for exclusion of bed-ridden and wheelchair-bound individuals)4
  • No history of bilateral hip fractures or hip replacement surgery (this exclusion is used generally in efficacy trials regarding hip protectors to assure a certain level of functional status)4
  • Must have at least one risk factor for hip fracture: history of prior fracture, physician-diagnosed osteoporosis or osteopenia, low BMI (≤20 (kg/m2)), family history of hip fracture or past/current smoker5(There is little evidence of benefit from hip protectors among lower risk elderly2)
  • Must not have ulcers, blisters or skin lesions around the hip area that would be covered by the SafeHip Soft hip protector4 (would make compliance difficult)
  • Must not have a waist circumference that exceeds that of the SafeHip Soft hip protector6

(so the participant is physically able to fit into the SafeHip Soft hip protector)

  • Must not have been recommended for exclusion by nursing home staff due to behavioral issues affecting adherence3 (to ensure individuals with a history of resistance to interventions or noncompliance are excluded)
  • Must be able to provide informed consent (to ensure participants are of functional mental status)

Sites:

Study to take place at nursing homes (NH)

Why nursing homes?

  • House a large, growing proportion of the elderly population: As of 2010, there were 1,385,251 residents in NHs throughout the US.7The majority of residents are over the age of 70.7
  • High risk for falls and hip fractures: Osteoporosis and osteoppenia, risk factors for hip fractures, are highly prevalent in NH facilities. One study found that 97% of residents have osteoporosis or osteopenia.1Studies suggest that elderly who are in NH facilities are at greater risk of hip fracture than elderly of the same age and sex who live independently.8 One study by Zimmerman and colleagues estimated the rate of hip fractures for NH residents to be anywhere from three to eleven times that of age-matched elderly living independently.1 Focusing on a population with high fall and fracture rates, will allow investigators to see if our hip protector works to reduce fractures.
  • Ease of screening, data collection, and study management:NHs by their nature offer benefits for running an evaluation study because the residents will be easier to track and adherence to protocols can be better monitored. This will help keep down costs and reduce time and resources needed to conduct the study. Assisted living facilities (ACLs) for the elderly could also be used, but these residents are not as at great a risk of falling or experiencing fractures.2This low incidence of events may make it challenging to show an effect from the SafeHip Soft protector. Although studying independent elderly living in the general community would be interesting, as they often have not experienced a major fall or fracture and may benefit from avoidance of such an incident, study set up and monitoring would require considerably more resources as participants are dispersed. Furthermore, as with ACLs, incidence of falls might be too low in order to find out if the hip protector prevents fractures.

Sampling Frame

All certified NH facilities within the large nursing home management network HCR ManorCare (Toledo, Ohio; over 38,000 beds in 277 facilities; In order to be eligible, facilities must have 30 residents or more and 80% of residents age 70 and older (these criteria have been used in prior studies to assure it is an elder-centered NH).3 NHs must not already be participating in a trial on hip protectors or have hip protector protocols in place (limits external validity threat of multiple treatment effect).

Sampling Strategy

Contact all eligible NHs in HCR ManorCare, inviting them to participate.

Those who agree to participate will go through an informed consent process for the site and any staff who will be working on the project.

Among consenting nursing homes, match by facility and patient-level characteristics: number of residents, demographics of residents (age, gender, race), number of resident falls in the past year, and resulting number of hip fractures, facility occupancy rate (number of total residents divided by the number of licensed beds) and staff-to-resident ratio (the number of staff members providing direct nursing care to residents divided by the total number of residents in the facility.Note: this should include all residents and staff, not merely those participating in the study).6

Within each matched pair, randomly assign each site to the intervention or the control group.

Note on Sample Size: Prior studies randomized at the NH facility level have estimated that around 500 participants in each group would allow detection of a reduction (one-sided hypothesis) of relative risk of hip fracture by 50% at alpha 0.05 and80% power.3This gives investigators and idea of the sample size that may be needed in the evaluation to show an effect. Final sample size requirements will be determined through power analysis once the evaluation procedure is finalized.

Justification for randomizing at the facility, rather than resident-level: Although randomizing at the facility level may introduce some logistical challenges such as need for a very thorough matching procedure to assure facilities are similar as possible and the need to have similar participation rates among NHs, there are several major concerns that arise with randomization at the resident-level. The concern with resident-level randomization is that residents will likely realize what group they are in and may react positively or negatively.9 In addition, randomization at the individual level may make it more difficult for the staff to keep track of who is to receive what intervention. Though facility-level randomization may pose more challenges to the setup of the intervention, it will streamline the intervention, make it simpler for facilities and staff to implement and avoid any internal validity threats such as contamination or Hawthorne effect from allowing control individuals to be exposed to intervention group individuals.

Independent Variables:

Whether and to what extent the resident follows the prescribed protocol for wearing the SafeHip Soft hip protector is the independent variable in this study.

Adherence is a major concern in hip protector studies.10 A review in 2002 by Schoor and colleagues estimated an average of 56% adherence, with a range from 20% to 92%.10 Given the potential for low adherence rates, the intervention should work to encourage compliance as well as thoroughly track adherence of participants.

Prescribed Protocol for wearing the device:Waking-Hours Use

  • Participants should be measured and fitted for the SafeHip Soft
  • Participants should be informed on hip health and the importance of compliance. They will also be shown how to wear the garment. Respondents will be expected to demonstrate knowledge of the protocol by putting the SafeHip Soft on themselves during the training. If the resident has difficulties putting on the SafeHip Soft in the demonstration, this will be noted by the study coordinator and staff will be informed of the potential need for assistance for that particular participant.
  • Note: Under the definition of adherence to be used in this study, compliance is not defined solely by a participant’s ability to put on the garment on their own. Though this is preferred, it is recognized that participants will have varying levels of physical ability and that some may need assistance with putting the protector on properly.
  • It is recommended that the intervention focus on use during daytime-hours. Compliance is likely to be higher during these hours and the high activity levels during these hours increase risks of falls and fractures.4Round-the-clock use of hip protectors is the standard for most research studies and is often recommended by physicians, particularly if the participant is getting up two or more times a night.11 However, most studies have shown particularly low compliance at night (roughly 3%).12

Definition of Compliance:

Wearing the SafeHip Soft protector continually and properly during daytime hours (6:00 to 24:00).

The definition of “properly” will be if the following criteria are met: SafeHip Soft is worn a) under the garments, b) with waist band at or around their waste and c) the pads of the hip protector are covering at least half of the hip area.

Measurement of Compliance:

NH care staff will collect compliance data on a standardized participant, specific diary sheet to record whether hip protectors were worn (and worn properly) during daytime hours

Daytime hours will be divided into three parts, as has been done in other studies: Morning (06:00–11:59), afternoon (12:00–17:59), evening (18:00–23:59). Overnight use will not be recorded (24:00 to 5:59).12

Care staff will record on the provided diary sheet whether the hip protector is being worn (including not only having it on, but having it on properly) during each time session.

Daytime compliance will be defined as the proportion of sessions where hip protectors were worn, divided by the total number of sessions (365 days of observation * 3 sessions per day = 1095 total sessions). This measurement technique will allow us to explore the possibility of a dose-response relationship between level of compliance and our outcomes of interest.

To verify compliance reports from staff, research staff will visit unannounced once a week, during various shifts and days of the week.4

Methods for Improving Compliance: Recommend a one month run-in period to exclude residents with poor adherence. A run-in procedure will help assure participants in the study are likely to be compliant. Although this might limit generalizability of findings to those who exhibit a certain level of compliance to the procedure (which we can directly measure through the compliance measure recommended prior to this section) it helps assure investigators can answer the one of the central questions, which is whether the SafeHip Soft is effective at preventing hip fractures.

Research staff will make four unannounced compliance visits over a two-week period post-participant training.

If during two of these occasions, the participant was not wearing the hip protector or not wearing it properly, they will be excluded from the study.4

Study Design:

1)Option 1: Randomized Control Trial, with a pre-test / multiple post-test design

R O1 X O2O3O4

R O5[ ] O6O7 O8

Design Details: NH facilities will be matched (see sampling strategy for matching details) into pairs prior to randomization.Matched pairs will then be randomized to the intervention or control group. Prior tointroduction of the intervention, a baseline survey would be conducted among participating residents assessing the major dependent variables of this study. In order to blind respondents to the main outcomes of the survey, these metrics will be included within a more general health and wellness survey. The experimental group would then be provided with the SafeHip Soft hip protectors and given basic information on how to use the protectors, hip health and importance of compliance. To assure the two groups stay as similar as possible, the control group could be provided with training on nutrition. The same survey delivered at baseline will be administered two weeks post-training, 6 months post-training and 1 year post-training to see changes in any of the outcome variables. Note: As the main outcome variables will be masked in a more general survey about health and wellness, control group members will likely not think the questions around falls and hip health will be out of the ordinary.

a)Strengths:

i)If randomization is successful, this study design will increase the likelihood that the intervention and control nursing homes are alike at baseline thus increasing internal validity.9 This helps rule out alternative explanations for any effects seen from the intervention. Successful randomization will rule out internal validity threats from: History, Maturation, Testing, Regression, Selection, and possible interactions between selection and other threats (history, maturation, testing and instrumentation).9

ii)Can use baseline pre-testresults to test whether randomization produced equivalent groups (benefit over a post-test only randomized control trial)

iii)Can use measures of change with a pre-test/post-test design. These measures are more statistically sensitive.9

iv)Uses the same survey measure at all time points, removing instrumentation threat to internal validity.9

b)Weaknesses:

i)Requires a large enough sample size in order to increase likelihood that randomization will work. If randomization fails, the nursing homes will not be similar at baseline opening up the study to more internal validity threats that may need to be controlled for in the analysis.

ii)Randomization does not eliminate all threats to internal validity. Potential threats to validity, even if randomization is successful include:

Attrition / If loss to follow up (due to illness, death or other causes) occurs differentially between the intervention and control group this could be an issue. Particularly when working with elderly, attrition is of particular concern.4
Hawthorne Effect / Even though the groups will be physically separate as the NHs will be collectively assigned to one treatment or the other. Those receiving the hip protector may change behavior just by the fact that they are being given something. By providing the controls with some sort of protocol (such as instruction regarding nutrition) or observation, this may help control for Hawthorne effect. However, it is still a major threat that could lead to:
  • Intervention group responding more positively simply due to greater levels of observation (giving the answers researchers are looking for, feeling special, etc.). This would lead to greater differences between the intervention and control group.
  • Resentful demoralization*: Controls become resentful or demoralized about being deprived of the hip protector intervention becoming less motivated and cooperative. This may lead to greater differences between the two groups
  • Compensatory rivalry*: Controls may respond by competing to outperform the intervention group. This would reduce the differences seen between the two groups.
*Note, these responses assume controls know something is being withheld. If informed consent procedures do not require disclosure that this is specifically an intervention trial surrounding hip protectors, the research may be able to avoid these threats.
Compensatory Equalization of Treatments / NH sites and staff are likely to realize whether or not they are receiving the intervention. If the informed consent process allows for investigators to remain vague about the distribution of hip protectors as the intervention (and instead can just mention that the study is regarding hip health and prevention techniques) then blinding may be possible. If not, or if NH sites have contact with one another, the control facilities may realize they are not receiving hip protectors. This realization could lead to them seeing fit to provide the intervention to their participants (perhaps by purchasing hip protectors and creating internal protocols encouraging use).

iii)Potential for external validity issues:

Testing-treatment interaction
(Potentially major threat) / Possibility that the pre-test actually becomes part of the intervention. There is no way to know if the same results would be seen if the pre-test were removed. The results may thus not be generalizable without the pre-test.
Selection-treatment interaction
(Potentially major threat) / Because this is limited to residents of nursing homes, there is a question of whether the results apply to other elderly individuals. Those in nursing homes tend to be more frail, prone to falls and have higher rates of risk factors for hip fractures.1
Setting-treatment interaction
(Potentially major threat) / Since the study would be run in a nursing home, question of whether would work for elderly living in other settings such as assisted living facilities, hospitals or those living independently. Particularly since those in nursing homes tend to get a high level of oversight and observation from staff whereas other living situations may have lower or higher levels of independence.
History-treatment interaction
(unlikely threat) / Whether the intervention would work in different time periods may not be a major threat as we are testing a device rather than trying to change attitudes and behaviors. However, if feelings toward compliance or ideas around efficacy of hip protectors were to change, this could affect compliance levels and may lead to different findings.

Note: External validity effect of multiple treatment effects is not an issue here because this has been controlled for in the sampling frame (NHs are excluded if they have existing programs or standardized protocols for hip protectors.