Abstract
Background
We compared the cost-effectiveness of 10 weeks of outreach rehabilitation (intervention) versus usual care (control) for ambulatory nursing home residents after hip fracture.
Methods
Enrollment occurred February 2011 through June 2015 in a Canadian metropolitan region. Seventy-seven participants were allocated in a 2:1 ratio to receive a 10-week rehabilitation program (intervention) or usual care (control) (46 intervention; 31 control). Using a payer perspective, we performed main and sensitivity analyses. Health outcome was measured by quality-adjusted life years (QALYs), using the EQ5D, completed at study entry, 3-, 6-, and 12-months. We obtained patient-specific data for outpatient visits, physician claims, and inpatient readmissions; the trial provided rehabilitation utilization/cost data. We estimated incremental cost and incremental effectiveness.
Results
Groups were similar at study entry; the mean age was 87.9 ± 6.6 years, 54 (71%) were female and 58 (75%) had severe cognitive impairment. EQ5D QALYs scores were nonsignificantly higher for intervention participants. Inpatient readmissions were two times higher among controls, with a cost difference of −$3,350/patient for intervention participants, offsetting the cost/intervention participant of $2,300 for the outreach rehabilitation. The adjusted incremental QALYs/patient difference was 0.024 favoring the intervention, with an incremental cost/patient of −$621 for intervention participants; these values were not statistically significant. A sensitivity analysis reinforced these findings, suggesting that the intervention was likely dominant.
Conclusion
A 10-week outreach rehabilitation intervention for nursing home residents who sustain a hip fracture may be cost-saving, through reduced postfracture hospital readmissions. These results support further work to evaluate postfracture rehabilitation for nursing home residents.
Keywords: Hip fracture, Nursing home, Rehabilitation, Cost-effectiveness, Functional recovery
Hip fractures are devastating injuries for older adults. Despite 20%–25% of these injuries occurring in residential care facilities (1), nursing home residents are infrequently included in research studies (2). Many of these residents have underlying cognitive impairment (3), a condition that also affects their eligibility for postfracture rehabilitation. The current evidence, although limited, suggests that these individuals experience poorer recovery from hip fractures and have a higher mortality rate (4–6). Evaluations of the cost of postfracture care in residential settings are limited (7,8). Heinrich and colleagues (9) described high health care costs in the first postfracture year, with additional costs related to readmissions and increased care needs for this vulnerable patient population. A recent randomized trial demonstrated short-term benefits of a 4-week rehabilitation program for similar patients, although these benefits were not sustained more than 12 months (10). This randomized trial also reported the program was not deemed cost-effective. In contrast, we recently reported a controlled feasibility trial that demonstrated modest benefits in mobility that were sustained to 12-month postfracture following a 10-week rehabilitation program (11). Our program was designed specifically for nursing home residents who were ambulatory prior to their fracture; there were no cognitive criteria for inclusion in the program. The intent of the current analysis was to perform a cost-effectiveness analysis on this rehabilitation program for nursing home residents.
Methods
Description of the Feasibility Trial
As outlined in a previous publication (11), eligible participants were ambulatory nursing home residents who sustained a hip fracture between February 2011 and June 2015 in a Canadian metropolitan health zone that provides universal health care coverage for 1.5 million residents. Non-English speaking residents and those who only ambulated from bed to wheelchair prefracture were excluded, and no cognitive criteria were applied for participation. Participants were assigned without knowledge of participants’ functional or cognitive status to the outreach intervention (intervention) or usual care (control) groups, at a 2:1 ratio as the outreach team could accommodate. We used proxy respondents to determine outcomes as 58 (75%) participants had severe cognitive impairment (ie, Mini Mental Status Exam Scores of <12 when assessed postoperatively before hospital discharge by a trained research associate). The outreach group received 10 weeks of rehabilitation following hospital discharge while the control group received usual care. The regional health ethics board provided research ethics approval (Pro00010006).
Economic Evaluation
This evaluation includes two cost-effectiveness analyses that were conducted from a public payer perspective (ie, the direct cost of health services incurred by the provincial ministry of health, Alberta Health [AH], and the single provincial health authority, Alberta Health Services [AHS]). The costs of all hospital inpatient, hospital outpatient and physician services utilized during the 1 year follow-up period (excluding the surgical hospital stay) were included. The cost of rehabilitation services comprising the outreach (intervention) program and usual care (control) was also included. Indirect costs, such as personal expenditures and productivity loss, of participants and caregivers were excluded. The primary analyses imputed values that were missing; a second sensitivity analysis based on available cases was also performed.
Since the analytic horizon was 1 year, there was no need to discount costs and QALYs (12). The analyses were conducted according to the intention-to-treat principle with the main analysis including 46 participants in the intervention and 30 participants in the control group. One control patient, with total health care costs 18 times the mean of other control participants, was deemed to be an outlier and was excluded from the analysis.
Health Outcomes
As recommended for cost-effectiveness analyses, the health outcome was measured by quality-adjusted life years (QALYs) (13), using the EQ5D-3L instrument. The EQ5D, used to measure general patient quality of life, is a validated nondisease-specific instrument based on econometric modeling (12). It has five dimensions representing attributes of personal health and activities of daily living. Proxy respondents were asked to complete the EQ5D questionnaire at baseline (prefracture), and at 3, 6, and 12 months. Although validated for use with older populations, including hip fracture, its use with proxy respondents is more limited (14,15); however, it appears that family members provide reasonable responses to less-observable domains (16). A single index score, with limiting values of 0 (dead) or 1 (no health problems), was derived from the results of each questionnaire by applying population-based utility weights (12). The index scores of the four observations were combined using the area under the curve method to derive patient-specific QALYs (17).
Cost Estimates
With recruitment occurring more than 4 years (2011–2015), the direct costs of physician and hospital services were adjusted to a single reference year (2015) in Canadian dollars (CAD). We obtained patient-specific utilization data for hospital outpatient clinics (including the emergency department) and hospital inpatient admissions from AHS, as well as physician claims records from AH. Data related to the utilization of rehabilitation services and payments to service providers were obtained from the feasibility trial.
We estimated hospital and physician costs at the service event level. Relative cost weights, based on patient characteristics and treatment variables, were assigned to hospital inpatient and outpatient service events that had been grouped to clinically relevant and statistically homogeneous categories (18). The cost estimates of hospital services were estimated as the product of the relative cost weight of each service and the Alberta cost per weighted case (19). These relative cost weights were based on grouping methodologies specific to inpatient and outpatient modes of care. Outpatient weights were scaled so that the inpatient cost per weighted case could be applied to outpatient services (20). The cost of operational overhead for each service was included in the estimates. AH uses a fee-for-service system to pay for physician services, based on a standard schedule. For all physician services, provided in hospital and in the long-term-care facilities, fees were adjusted to the 2015 schedule (21).
Outreach Intervention
Outreach participants received 30 sessions of rehabilitation more than 10 weeks in their nursing home after hospital discharge. Outreach rehabilitation teams consisted of a licensed physical therapist (PT) and two physical therapy assistants (PTAs) who were hired and trained by the investigators to provide the rehabilitation program. Usual rehabilitation services that would normally be provided by the long-term-care facility were discontinued for the intervention participants during the 10-week intervention period. The cost of the intervention was estimated as the product of the number of weeks each patient received the intervention and the cost per week. In order to estimate the weekly unit cost, provider-specific payments records were accessed reflecting salary, benefits, and operational overhead. Since the intervention services were delivered by nonfacility providers, travel cost between the research office and the long-term-care facilities was also included.
Control (usual care)
Control participants received usual postfracture care in their nursing home after hospital discharge. Nursing home personnel logged their hours of rehabilitation care for control participants, but received no directions for postfracture rehabilitation care. Complete rehabilitation care data were available for only 9 of 30 controls, with the remaining participants providing incomplete or no rehabilitation information. Two control participants who survived less than a week were assumed to have no rehabilitation cost. In keeping with the outreach intervention providers, control rehabilitation costs included salary, benefits, and overhead.
All other services provided by the long-term care facilities to both intervention and control participants were not included patient-specific costs in long-term-care facilities not reported in Alberta. Nevertheless, we expected that services provided to all participants by the facilities would be virtually identical after the intervention period between the two groups.
Missing Data and Imputation
About 9% of the data cells required to estimate incremental cost and QALYs were missing from the study data set. Missing data occurred in only quality-of-life (EQ5D) variables and the rehabilitation cost for the control group. The proportion of missing quality-of-life data was similar between groups: 8% and 11% for the intervention and control groups, respectively. The missing values were distributed more than 44 participants, so the complete case analysis was based on 32 (42%) participants. A smaller complete case number lowers the statistical power of the analysis, increasing the risk of biased estimates (22).
Since data were collected from proxy respondents, we assumed that missing data were not related to adverse outcomes and were generally “missing at random.” (22) A multiple imputation procedure was undertaken at the patient level to estimate missing values. Twenty data sets were generated to incorporate the uncertainty involved in the imputation process (23). We used the Amelia II v1.7.3 package for R to conduct multiple imputation analysis, using the observed data and the empirical relationships between variables to estimate the missing values (24).
Cost-Effectiveness Analysis
We assessed cost-effectiveness by estimating incremental cost (the difference in mean cost per patient between the study arms) and incremental effectiveness (the difference in mean QALYs per patient). An intervention is considered cost-effective, in relation to an alternative, in the following situations: (a) it costs less (incremental cost is negative) and is more effective (the incremental health effect is positive); (b) the intervention costs more and is more effective than the alternative, but society is willing to pay for the additional cost per QALY; and, (c) theoretically at least, the intervention costs less and is less effective, but society is unwilling to pay for the additional cost per QALY of the alternative (12,25,26). Interventions that cost less and are more effective are considered dominant or “cost-savings.”
We used regression analysis to estimate incremental cost effect, adjusting for the length-of-stay (LOS) on the index hospital event and survival time during the follow-up period. The baseline EQ5D score was the only variable used to adjust the incremental health effect. The Rcmdr package v2.1–5 for R was used to conduct the analyses (27) and Filemaker Pro Advanced v16.05 (Filemaker, Inc., Santa Clara) was used for general database management.
Uncertainty
To account for the uncertainty due to sampling variation, and to overcome the inherent skewness in sampling distributions of economic variables, we used a nonparametric bootstrap analysis to generate the joint distribution of incremental cost and health effect (12,26,28). We used the boot package v1.3–20 for R to generate 10,000 replications for each of the 20 imputed data sets (29). The bootstrap replicates were used to derive a cost-effectiveness acceptability curve (CEAC), indicating the probability of the intervention being cost-effective at various levels of society’s willingness to pay per unit of health effect gained (30,31). Confidence intervals, for incremental cost and health effect estimates, were based on estimates of standard errors obtained by the application of Rubin’s method to the bootstrap replicates of the imputed data sets (32).
Sensitivity Analysis
To further assess the influence of missing data, we conducted a cost-effectiveness analysis based on available cases: 57 participants had complete cost data and 44 complete EQ5D data. Regression analyses were used to estimate the incremental cost and health effect. Since the number of complete cases for cost and health outcome data were different, we were not able to bootstrap both the cost and QALY regressions simultaneously. In order to estimate the CEAC, we employed an alternative approach based on the application of the central limit theorem (ie, the sampling distribution of an estimator will tend to normality as the sample size increases) (14). In order to estimate confidence intervals for incremental cost and QALYs, we used a separate bootstrap analysis (10,000 replications) for each regression.
Results
As previously reported, 77 participants were enrolled in the trial, with 46 allocated to the intervention group and 31 to control (with one outlier removed for a total of 30 for this evaluation) (11). After removing the patient deemed to be an outlier, there were no significant group differences at study entry, aside from a slight delay in hospital rehabilitation starts postoperatively for the intervention group (Table 1). The mean age of the cohort was 87.9 ± 6.6 years, 54 (71%) were female and 58 (75%) had severe cognitive impairment as measured by the Mini-Mental Status Examination (ie, scores <13) (15). Both groups reported high levels of dependence prefracture based on Functional Independence Measure Motor domain (FIMmotor) scores, but 42% of participants were independently walking (ie, FIM ambulation scores >5; p = .91 for group differences). The EQ5D baseline scores based on the imputed data set (data missing for seven participants) were not significantly different (95% CI, Table 2), although the mean score for intervention participants was 0.08 higher than control participants.
Table 1.
Clinical Data of REGAIN II Participants at Study Entry by Group Allocation
| Mean ± SD or n (%) | |||
|---|---|---|---|
| Clinical Variable | Intervention (n = 46) | Control (n = 30) | p Value |
| Participant characteristics | |||
| Age in years | .78 | ||
| <85 | 10 (22) | 8 (27) | |
| 85+ | 36 (78) | 22 (73) | |
| Gender, female | 31 (71) | 23 (77) | .45 |
| Comorbidities | .38 | ||
| ≤2 | 11 (24) | 4 (13) | |
| 3 or more | 35 (76) | 25 (86) | |
| Walker type | 1.0 | ||
| Outdoor walker | 13 (30) | 8 (28) | |
| Indoor walker | 31 (70) | 20 (72) | |
| Assistance | .31 | ||
| No/minimum | 37 (88) | 20 (77) | |
| Moderate to maximum | 5 (12) | 6 (23) | |
| Proxy | 1.0 | ||
| Family member (spouse/offspring) | 39 (89) | 25 (89) | |
| Other | 5 (11) | 3 (11) | |
| Functional independence measure (FIM) | |||
| FIMmotor domain | 43.8 ± 18.4 | 39.9 ± 15.6 | .35 |
| Locomotion domain | 6.1 ± 2.7 | 5.6 ± 2.3 | .39 |
| Mobility domain | 11.2 ± 4.7 | 10.9 ± 4.5 | .75 |
| Surgical and hospital data | |||
| Fracture type | .33 | ||
| Femoral neck | 15 (33) | 14 (47) | |
| Trochanteric | 30 (67) | 16 (53) | |
| Hospital complications | 1.0 | ||
| Yes (1or 2 complications) | 7 (15) | 5 (17) | |
| Hospital days | .65 | ||
| Median (IQR) | 7 (5–10) | 8 (6–11) | |
| [range] | [3–18] | [3–18] | |
| Surgery days | .73 | ||
| Median (IQR) | 1 (0–1) | 1 (0.75–1.25) | |
| [range] | [0–4] | [0–3] | |
| Postoperative rehabilitation start days | .006 | ||
| Median (IQR) | 1.5 (1–2) | 1 (1–1.5) | |
| [range] | [1–3] | [1–4] |
Note: SD = standard deviation; FIM = functional independence measure; IQR = interquartile range.
Table 2.
Health Outcome by Group Allocation*
| Variable/Observation | Intervention n = 46 | Control n = 30 | Mean Difference [95% CI]† |
|---|---|---|---|
| EQ5D baseline | 0.558 | 0.477 | 0.081 [−0.037, 0.199] |
| EQ5D 3 month | 0.283 | 0.235 | 0.048 [−0.045, 0.141] |
| EQ5D 6 month | 0.259 | 0.214 | 0.045 [−0.061, 0.151] |
| EQ5D 12 month | 0.235 | 0.180 | 0.055 [−0.055, 0.165] |
| QALY l2 month | 0.297 | 0.244 | 0.053 [−0.038, 0.148] |
Note: EQ-5D = Euro Quality of Life 5 Dimension; QALYs = Quality Adjusted Life Years; CI = Confidence Intervals.
*Based on multiple imputation: 20 imputed data sets.
†Unadjusted for baseline variables.
Health Outcomes
At all observation points after baseline, the EQ5D scores were higher for intervention participants than controls, but these differences were not significant (Table 2). Intervention participants experienced a nonsignificant difference of 0.051 QALYs per participant in relation to controls based on imputed data that was unadjusted for baseline differences between participants.
Health Service Utilization and Costs
Hospital outpatient visits and physician services per participant were virtually equal between study arms (Table 3). However, the control group had a higher proportion of inpatient readmissions in the 12-month follow-up period relative to the intervention group; this finding was not statistically significant. Overall, there were 14 readmissions in 13 patients (8 readmissions in 7 (23%) of 30 patients in the control group vs 6 (13%) of 46 patients in the intervention group). Readmissions appeared related to the hip fracture event with five readmissions associated with infectious events (septicemia [n = 3 control] or urinary tract infections [n = 2 intervention]), five related to issues with fracture fixation / mobility issues (four control; one intervention), and two related to new fracture events (two intervention). The remaining two readmissions were associated with renal failure (1 control) and mental health (1 intervention).
Table 3.
Cost and Utilization per Patient by Group Allocation*
| Cost Component | Intervention n = 46 | Control n = 30 | Difference [95% CI]† |
|---|---|---|---|
| Intervention/rehabilitation‡ | 2,741 | 441 | 2,300 [1,688, 2,912] |
| Inpatient | |||
| Cost per patient | 1,598 | 4,947 | −3,350 [−7,649, 949] |
| Admissions per patient | 0.13 | 0.27 | −0.14 [−1.38, 0.17] |
| Hospital outpatient | |||
| Cost per patient | 628 | 814 | −185 [−668, 297] |
| Visits per patient | 1.6 | 1.7 | −1.0 [−0.9, 0.6] |
| Physicians | |||
| Cost per patient | 2,080 | 2,156 | −76 [−849, 698] |
| Services per patient | 61 | 59 | 2 [−16, 21] |
| Total cost per patient* | 7,047 | 8,359 | −1,312 [−6,187, 3,563] |
Note: CI = confidence interval.
*Costs are based on 2015 Canadian Dollars.
†Unadjusted for baseline variables.
‡Based on multiple imputation: 20 imputed data sets.
The cost difference for inpatient readmissions was $3,350 CAD less/participant for the intervention than the control group. This readmission cost more than offset the estimated excess cost per participant of $2,300 CAD for the outreach group relative to the control group. Consequently, the unadjusted total cost per participant was $1,312 CAD less for the intervention group; again this finding was not statistically significant.
Cost-Effectiveness Analysis
After adjusting for baseline differences among participants, the incremental QALYs per participant declined to a difference of 0.024 (Table 4) in favor of the intervention. The adjusted incremental total cost per participant was $621 CAD less for the intervention group. Although both incremental values were not statistically significant, the point estimates of this main analysis, based on multiple imputation, did indicate dominance (less costly and more effective) of the intervention versus control (usual care). The CEAC shown by the solid line in Figure 1, embodying the uncertainty of the estimation process, shows that the intervention was more likely to be cost-effective than the control. The CEAC rises from a 55% probability of the intervention being cost-effective at the origin, where society will not pay any premium for additional costs per QALY to 70% at the final point on the curve where society’s willingness-to-pay is $100,000 per QALY gained per participant.
Table 4.
Final CEA Results* by Group Allocation†
| Analyses | Intervention | n | Control | N | Unadjusted Mean Difference | Adjusted Difference‡ [95% CI]§ |
|---|---|---|---|---|---|---|
| Main analysis* | ||||||
| Cost per patient | 7,047 | 46 | 8,359 | 30 | −1,312 | −621 [−6,546, 5,165] |
| QALYs per patient | 0.297 | 46 | 0.244 | 30 | 0.053 | 0.024 [−0.058, 0.108] |
| Sensitivity analysis | ||||||
| Cost per patient | 7,047 | 46 | 10,590 | 11 | −3,543 | −1,407 [−12,122, 4,944] |
| QALYs per patient | 0.328 | 24 | 0.210 | 20 | 0.118 | 0.068 [−0.033, 0.192] |
Note: CEA = cost effectiveness analysis; QALYs = quality adjusted life years; CI = confidence intervals.
*Costs are based on 2015 Canadian Dollars.
†Based on multiple imputation: 20 imputed data sets.
‡Adjusted for baseline variables.
§Based on 10,000 bootstrap replicates of each imputed data set for the main analysis and 10,000 replicates of each incremental cost and QALYs gained.
Figure 1.
Cost-effectiveness acceptability curves. Note: Costs are expressed in constant 2015 Canadian dollars.
Sensitivity Analysis
The results of the available case analysis reinforce the results of the main analysis, indicating that the intervention was likely to be dominant. The nonsignificant incremental cost and health effect point estimates were higher than the main analysis. However, the width of CIs was greater for corresponding incremental values in the available case analysis compared with those of the main analysis, indicating greater uncertainty. Nevertheless, the CEAC shown by the dotted line is higher than the curve for the main analysis, reinforcing the conclusion that the rehabilitation intervention may be cost-effective compared to usual care (Figure 1).
Discussion
Evidence regarding care and outcome of nursing home residents who sustain a hip fracture is limited, with little known about the costs of providing postfracture care for these individuals. We recently demonstrated that participation in a standardized rehabilitation program was feasible, and provided modest benefits in mobility that were maintained to 12-month postfracture (11). The current analysis suggests that a prolonged rehabilitation program (10 weeks) may also be cost-saving from a health care payer perspective, primarily due to a reduction in postfracture hospital readmissions. The group that received the rehabilitation program had postfracture health care costs that were $621 CAD less than those who received usual care over the first 12 postfracture months despite the $2,300 CAD costs per participant associated with providing the rehabilitation program. The cost-effectiveness findings, although promising, were not statistically significant as they were obtained from a feasibility trial that was not powered for a full effectiveness or cost-effectiveness analysis.
However, the findings are encouraging, as rehospitalizations are common after hip fracture for people living in nursing homes. Mitchell and colleagues (16) reported that aged care residents in Australia had fewer rehabilitation episodes than those who were community-dwelling after their hip fracture and were more likely to have a hospital readmission. Heinrich and colleagues (9) reported increased postfracture costs associated with both rehabilitation and readmissions after hip fracture for nursing home residents in Germany, but did not perform a cost-effectiveness analysis to examine the impact of postfracture care and determine the relative benefit of rehabilitation in preventing or avoiding future complications requiring hospital readmission.
While our feasibility trial indicates a modest, but sustained mobility benefit after a 10-week post hip fracture rehabilitation program for nursing home residents, another well-powered randomized trial did not find sustained benefits after a 4-week rehabilitation program for a similar group of residents (10). In contrast to our current cost-effectiveness findings, the 4-week program in that randomized trial was not cost-effective. That trial and ours represent the most rigorous evaluations of postfracture rehabilitation and its associated costs for nursing home residents (7,8). Despite contradictory findings, the difference in the interventions evaluated in these trials, predominantly the duration of the rehabilitation program, would suggest that further research is warranted for this vulnerable patient group. It is possible that postfracture rehabilitation needs to be extended for a longer time period to allow these frail individuals to achieve modest but sustained benefits. Although extended rehabilitation adds cost, sustained benefits of the program may offset the costs associated with rehabilitation by avoiding future hospital readmissions. Avoiding hospital readmissions has both payer and patient benefits.
The cost-effectiveness analysis performed herein used accepted methodology to determine the cost and impact associated with the intervention. However, there are some notable limitations. Although we had complete cost data on the intervention program, the usual care (control) group yielded very limited data, despite repeated and sustained efforts by the investigators to obtain usual practice data from the participating nursing homes. Only 9 of 30 health care providers for the usual care group returned a fully completed rehabilitation log, so we had to use imputation to build our comparison model. In our previous qualitative evaluation, caregivers in these settings acknowledged that it was difficult to provide rehabilitation and care after hip fracture due to a lack of resources (both personnel and time) (33,34). It was not possible, based on our data to know why rehabilitation services were so limited in the control group and whether rehabilitation treatment decisions were based on resource limitations or patient capacity to participate.
Finally, despite applying accepted analytic approaches, our evaluation was under-powered as it was performed with data obtained in a feasibility trial, which did not use a power calculation, so our results must be considered preliminary. Despite these limitations, this evaluation provides valuable information for others considering rehabilitation for nursing home residents after a hip fracture.
In summary, a 10-week outreach rehabilitation intervention for ambulatory nursing home who sustain a hip fracture, not only provides modest gains in mobility that were sustained to 12-month postfracture, but may also be cost-effective (ie, costs less, but is more effective). The cost-saving appears to arise from a reduction in postfracture hospital readmissions. These results should be considered preliminary, but they support further work evaluating the impact and value of postfracture rehabilitation for this most vulnerable group of patients sustaining hip fracture.
Acknowledgments
This study is based in part on data provided by Alberta Health. The interpretation and conclusions contained herein are those of the researchers and do not necessarily represent the views of the Government of Alberta. Neither the Government nor Alberta Health express any opinion in relation to this study.
Funding
This work was supported by Alberta Innovates Health Solutions Population Health Investigator Establishment Grant (RES0000396).
Conflict of Interest
L.A.B. is the David Magee Endowed Chair in Musculoskeletal Research and receives salary support from the Faculty of Rehabilitation Medicine at the University of Alberta. J.S.M. reports grants from the National Institute on Aging during the conduct of the study; he also received advisory board or consulting fees while conducting and preparing the paper for this study from: Ammonett, American Orthopaedic Association, Eli Lilly, Novartis, Pluristem, Sanofi, Scholar Rock, Viking, and a grant from Eli Lilly. S.R.M. held the Endowed Research Chair in Patient Health Management supported by the Faculties of Medicine and Dentistry and Pharmacy and Pharmaceutical Sciences at the University of Alberta. All other authors (D.L., C.A.J., D.W.C.J., D.M.W.) declare no conflicts of interest.
References
- 1. Berry SD, Lee Y, Zullo AR, Kiel DP, Dosa D, Mor V. Incidence of Hip Fracture in U.S. Nursing Homes. J Gerontol A Biol Sci Med Sci. 2016;71:1230–1234. doi: 10.1093/gerona/glw034 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Mundi S, Chaudhry H, Bhandari M. Systematic review on the inclusion of patients with cognitive impairment in hip fracture trials: a missed opportunity? Can J Surg. 2014;57:E141–E145. doi: 10.1503/cjs.023413 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Jones CA, Jhangri GS, Feeny DH, Beaupre LA. Cognitive status at hospital admission: postoperative trajectory of functional recovery for hip fracture. J Gerontol A Biol Sci Med Sci. 2017;72:61–67. doi: 10.1093/gerona/glv138 [DOI] [PubMed] [Google Scholar]
- 4. Seitz DP, Gill SS, Austin PC, et al. Rehabilitation of older adults with dementia after hip fracture. J Am Geriatr Soc. 2016;64(1):47–54. doi: 10.1111/jgs.13881 [DOI] [PubMed] [Google Scholar]
- 5. Crotty M, Miller M, Whitehead C, Krishnan J, Hearn T. Hip fracture treatments—what happens to patients from residential care? J Qual Clin Pract. 2000;20:167–170. doi: 10.1046/j.1440-1762.2000.00385.x [DOI] [PubMed] [Google Scholar]
- 6. Beaupre LA, Cinats JG, Jones CA, et al. Does functional recovery in elderly hip fracture patients differ between patients admitted from long-term care and the community? J Gerontol A Biol Sci Med Sci. 2007;62:1127–1133. doi: 10.1093/gerona/62.10.1127 [DOI] [PubMed] [Google Scholar]
- 7. Bulamu NB, Kaambwa B, Ratcliffe J. Economic evaluations in community aged care: a systematic review. BMC Health Serv Res. 2018;18:967. doi: 10.1186/s12913-018-3785-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Easton T, Milte R, Crotty M, Ratcliffe J. Where’s the evidence? A systematic review of economic analyses of residential aged care infrastructure. BMC Health Serv Res. 2017;17:226. doi: 10.1186/s12913-017-2165-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Heinrich S, Rapp K, Rissmann U, Becker C, König HH. Service use and costs of incident femoral fractures in nursing home residents in Germany: the Bavarian Fall and Fracture Prevention Project (BF2P2). J Am Med Dir Assoc. 2011;12:459–466. doi: 10.1016/j.jamda.2010.11.008 [DOI] [PubMed] [Google Scholar]
- 10. Crotty M, Killington M, Liu E, et al. Should we provide outreach rehabilitation to very old people living in Nursing Care Facilities after a hip fracture? A randomised controlled trial. Age Ageing. 2019;48:373–380. doi: 10.1093/ageing/afz005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Beaupre LA, Magaziner JS, Jones CA, et al. Rehabilitation after hip fracture for nursing home residents: a controlled feasibility trial. J Gerontol A Biol Sci Med Sci. 2019;74:1518–1525. doi: 10.1093/gerona/glz031 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Drummond M, Sculpher M, Torrence G, O’Brien B, Stoddard G.. Methods for the Economic Evaluation of Health Care Programmes. 3rd ed. Oxford: Oxford University Press; 2005. [Google Scholar]
- 13. Ramsey SD, Willke RJ, Glick H, et al. Cost-effectiveness analysis alongside clinical trials II-An ISPOR Good Research Practices Task Force report. Value Health. 2015;18:161–172. doi: 10.1016/j.jval.2015.02.001 [DOI] [PubMed] [Google Scholar]
- 14. Nixon RM, Wonderling D, Grieve RD. Non-parametric methods for cost-effectiveness analysis: the central limit theorem and the bootstrap compared. Health Econ. 2010;19:316–333. doi: 10.1002/hec.1477 [DOI] [PubMed] [Google Scholar]
- 15. Folstein MF, Folstein SE, McHugh PR. "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–198. doi: 10.1016/0022-3956(75)90026-6 [DOI] [PubMed] [Google Scholar]
- 16. Mitchell R, Draper B, Harvey L, Wadolowski M, Brodaty H, Close J. Comparison of hospitalised trends, treatment cost and health outcomes of fall-related hip fracture for people aged ≥65 years living in residential aged care and the community. Osteoporos Int. 2019;30(2):311–321. doi: 10.1007/s00198-018-4800-6 [DOI] [PubMed] [Google Scholar]
- 17. Manca A, Hawkins N, Sculpher MJ. Estimating mean QALYs in trial-based cost-effectiveness analysis: the importance of controlling for baseline utility. Health Econ. 2005;14:487–496. doi: 10.1002/hec.944 [DOI] [PubMed] [Google Scholar]
- 18. Canadian Institute for Health Information. Cost of a Standard Hospital Stay: Appendices to Indicator Library—Methodology Notes. Ottawa, Canada: Canadian Institute of Health Information; 2016. [Google Scholar]
- 19. Canadian Institute for Health Information. Cost of a Standard Hospital Stay 2016. https://yourhealthsystem.cihi.ca/inbrief/?lang=en#!/indicators/015/cost-of-a-standard-hospital-stay/;mapC1;mapLevel2. Accessed May 5, 2018.
- 20. Canadian Institute for Health Information. CACS Methodology. Toronto, Canada; 2012. [Google Scholar]
- 21. Alberta Health Care Insurance Plan 2015. Schedule of Medical Benefits (Procedures and Price List) as of April 1, 2015. Edmonton, Canada. [Google Scholar]
- 22. Schafer JL, Graham JW. Missing data: our view of the state of the art. Psychol Methods. 2002;7:147–177. [PubMed] [Google Scholar]
- 23. Graham JW, Olchowski AE, Gilreath TD. How many imputations are really needed? Some practical clarifications of multiple imputation theory. Prev Sci. 2007;8:206–213. doi: 10.1007/s11121-007-0070-9 [DOI] [PubMed] [Google Scholar]
- 24. Honaker J, King G, Blackwell J. Amelia II: A Program for Missing Data 2009. http://gking.harvard.edu/Amelia. Accessed September 15, 2016
- 25. Lier D, Robert M, Tang S, Ross S. Surgical treatment of stress urinary incontinence-trans-obturator tape compared with tension-free vaginal tape-5-year follow up: an economic evaluation. BJOG. 2017;124:1431–1439. doi: 10.1111/1471-0528.14227 [DOI] [PubMed] [Google Scholar]
- 26. Manca A, Sculpher MJ, Ward K, Hilton P. A cost-utility analysis of tension-free vaginal tape versus colposuspension for primary urodynamic stress incontinence. BJOG. 2003;110:255–262. [PubMed] [Google Scholar]
- 27. Fox J. The R Commander: a basic statistics graphical user interface to R. J Stat Softw. 2005;14(9):1–42. [Google Scholar]
- 28. Manca A, Palmer S. Handling missing data in patient-level cost-effectiveness analysis alongside randomised clinical trials. Appl Health Econ Health Policy. 2005;4:65–75. doi: 10.2165/00148365-200504020-00001 [DOI] [PubMed] [Google Scholar]
- 29.Canty R, Ripley B.https://cran.r-project.org Package “Boot.”. Accessed May 5, 2018.
- 30. Briggs AH, O’Brien BJ, Blackhouse G. Thinking outside the box: recent advances in the analysis and presentation of uncertainty in cost-effectiveness studies. Annu Rev Public Health. 2002;23:377–401. doi: 10.1146/annurev.publhealth.23.100901.140534 [DOI] [PubMed] [Google Scholar]
- 31. Fenwick E, Claxton K, Sculpher M. Representing uncertainty: the role of cost-effectiveness acceptability curves. Health Econ. 2001;10:779–787. doi: 10.1002/hec.635 [DOI] [PubMed] [Google Scholar]
- 32. Rubin DB. Multiple Imputation for Nonresponse in Surveys. Hoboken, NJ: Wiley Classics Library; 2004. [Google Scholar]
- 33. Wilson DM, Robertson SL, Jones CA, Johnston DWC, Beaupre LA. Barriers and facilitators to an outreach rehabilitation program delivered in nursing homes after hip fracture surgical repair. Adv Aging Res 2013;2(1):43–50. doi: 10.4236/aar.2013.21006 [DOI] [Google Scholar]
- 34. Buddingh S, Liang J, Allen J, Koziak A, Buckingham J, Beaupre LA. Rehabilitation for long-term care residents following hip fracture: a survey of reported rehabilitation practices and perceived barriers to delivery of care. J Geriatr Phys Ther. 2013;36:39–46. doi: 10.1519/JPT.0b013e3182569b4f [DOI] [PubMed] [Google Scholar]