Reducing mortality in hip fracture patients using a perioperative approach and “Patient- Centered Medical Home” model: a prospective cohort study

Hip fractures remain a major challenge to health care systems throughout the world [1], as both the number of hip fracture patients and the costs associated with their care increase over time [2‐6]. Despite improvements in implant technology, morbidity and mortality rates in the first postoperative year remain high [7]. Numerous reports have shown that perioperative approaches that include medical optimization, multidisciplinary care and rehabilitation decrease immediate perioperative morbidity [8]. However, the significant morbidities and costs associated with hip fractures more often occur after discharge and are related to long-term management of medical comorbidities [9]. Further complicating matters, it is difficult to distinguish between morbidity related to the hip fracture itself as opposed to the overall frailty and vulnerability of the elderly population [10]. In theory, an optimal approach to managing patients would initially optimize the patient for surgery, match implants to the patients’ functional condition, and subsequently manage the patient’s overall health status including existing comorbidities. This multimodal approach has not been previously reported in the orthopaedic literature. If this approach successfully decreased readmissions and mortality, any increased costs associated with more aggressive post-operative management might be offset within a health system [11].

At the same time that hip fractures challenge the orthopaedic community, there are recognized shortcomings in hospital post-discharge transitions of care in general. As recently as 2009, half of Medicare fee-for-service patients who had a 30-day readmission did not visit their physician before being readmitted [12]. Communication gaps are suspected to account for a significant share of readmissions, and these occur between providers (e.g., discontinued prescriptions), between provider and patient (e.g., lack of patient education) and in transitions between the hospital and community setting [13‐15]. Systematic reviews have reported significant reductions in 30-day readmission rates and cost savings associated with several enhanced discharge processes, most of which used enhanced in-hospital communication, payment incentives, and other multimodal approaches [16‐18].

The Patient-Centered Medical Home (PCMH) model, which evolved from the Chronic Care Model [19], involves enhancing the role of primary care providers as the locus of care coordination. The PCMH is not a physical location, but a redesign of healthcare delivery to facilitate partnerships among an individual patient and his/her personal physician and family, all supported by resources including enhanced information systems and communications infrastructure. The American Academy of Pediatrics (AAP), American Academy of Family Physicians (AAFP), and other societies representing approximately 330,000 U.S. physicians describe 7 joint principles of the PCMH. These principles include each patient having an ongoing relationship with his/her personal physician, highly coordinated care, enhanced access to care, and payment structures that reflect the value of care management work [20]. At our health system, a PCMH model attempts to reduce post-discharge morbidity by using nurse case managers to coordinate the transition from hospital to home. Case managers use early telephone outreach, medication reconciliation, social support assistance, and ensure timely follow-up with primary care physicians in the month immediately following surgery.

We compared rates of mortality, hospitalizations, emergency department visits and prescription orders between two prospective cohorts of hip fracture patients, both managed with identical perioperative protocols and one group subsequently managed via PCMH. Costs, functional outcomes and quality of life were also studied as secondary outcomes in a smaller subsample. We hypothesized that the best outcomes would occur in patients managed initially with a standardized approach and subsequently with PCMH management.

A total of 303 hip fracture patients met eligibility criteria for the study during the enrollment period, 108 of whom received PCMH post-discharge care and 195 of whom did not. The mean age was 81 years (range 60–100), and two-thirds (201 of 303) of patients were female. The overall survival rate at 12 months for this entire population was 219/303 (72%). After matching cases with controls, eleven PCMH patients had to be excluded for insufficient matches, resulting in a final study cohort of 97 PCMH cases matched to 97 non-PCMH controls. In the matched set, the mean age was 82 years (range 76–89) and 142/194 (73%) were female. Table 3 presents all baseline characteristics of the two primary matched cohorts (n = 194) for comparison.

For the secondary analysis of healthcare costs, claims data were available for 66/108 (61%) of the PCMH patients and 42/195 (21%) of the non-PCMH patients. For the quality-of-life and pain/function telephone questionnaires administered to non-deceased patients at 12 months, the survey response rate was 41/82 (50%) for PCMH patients and 50/137 (37%) for non-PCMH patients. Repeating the matching procedure for the analysis of secondary outcomes yielded 15 matched pairs with cost data who survived 12 months (n = 30), and 35 matched pairs with pain/function data who survived 12 months (n = 70).

Table 4 presents a summary of all outcomes at 6 and 12 months after surgery. At 6 months post-operatively, patients receiving PCMH post-discharge management had a significantly lower mortality rate than patients receiving standard care (11 vs. 26% respectively, p < 0.01). At 12 months, a difference persisted (23 vs. 30%, p = 0.12), though it was no longer statistically significant. Figure 1 presents the all-cause mortality survival curves for these two cohorts. The mean Harris Hip Pain/Function Score was also significantly improved at 12 months in the PCMH group as compared to controls (73 vs. 64, p = 0.04). Otherwise, no differences between the cohorts reached statistical significance. Differences in numbers of all-cause hospitalizations, ED visits and prescription orders per patient were not significantly different at 6 months (p = 0.69-0.91) or 12 months (p = 0.16-0.83). The mean EQ-5D quality-of-life index was similar between the two groups (0.73 vs. 0.76, p = 0.49), as was the mean EQ-5D Scale. For those patients surviving 6 months, mean pharmacy and non-pharmacy costs were lower in the PCMH cohort but not significantly so (p = 0.27-0.95), and a similar trend persisted at 12 months (p = 0.20-0.45).

Our study demonstrates, in a prospective study of patients receiving a unified in-hospital protocol, that the addition of post-discharge care from a “Patient-Centered Medical Home” program, including a nurse case manager for primary care-based follow-up, was successful in significantly reducing mortality in the 6 months following hip fracture and showed evidence of benefits to hip pain and function as well. While it may be unrealistic to expect that such a post-discharge program would eliminate mortality/morbidity, this study makes a promising case for utilizing PCMH programs for a focused, limited period following discharge. Although significant benefits in reduced admissions, emergency department visits or medication utilization and costs were not seen, the two cohorts remained similar in these respects, suggesting that the aggressive follow-up by the PCMH program did not substantially drive up healthcare utilization or costs. Future larger-scale studies should help to better define the optimal patient population for such programs and refinements that would provide further benefits for patients undergoing hip fracture surgery. We note that, while we found more published estimates of 1-year mortality than 6-month mortality, our control group's 6-month and 12-month mortality rates of 26 and 30%, respectively, appear to be consistently in the range of the published findings of Johnston et al. [29] and others who have followed operative treatment of hip fractures.

Mortality and morbidity after hip fracture in elderly patients remains a serious problem; however, there is growing evidence to suggest that patient factors and specific details of the approach to post-fracture treatment can impact mortality and quality of life. As noted by Johnston et al., many studies report mortality following hip fracture as a simple percentage and do not take into account age, gender and other factors [29]. In a recent systematic review, Butler et al. concluded that while neither implant type nor surgical approach were associated with differences in mortality rates, patient factors such as age, sex, pre-fracture functioning, and cognitive impairment were directly related to mortality and functional outcomes [30]. This literature suggested that the main mortality risk for elderly women is in the 6 weeks following surgery, while the risk for men is higher near 6 months, with comorbidities and age impacting different subgroups. In a study that complements our current results in a different population, Rahme et al. retrospectively reviewed 11,326 patients receiving hemiarthroplasty for hip fracture and showed that the small percentage of those (16%) who received post-discharge home care had a lower mortality rate at 3 months than those who received no home care [31]. While the data collected for this study do not allow us to pinpoint the exact reason why the PCMH program influenced the 6-month mortality rate in our study, we believe that by providing reinforcement during the critical 4 weeks following discharge, the primary care PCMH team may be able to better engage, activate and educate patients as well as identify and resolve issues that may otherwise be missed or caused by care gaps during the transition from hospital to home.

The main strengths of our study are its population and statistical design. We were able to use patients in an integrated health system where electronic health records were available for the hip fracture population to quantify mortality, inpatient admissions, emergency department visits and other healthcare utilization. Furthermore, we were able to take advantage of the fact that a PCMH program was in the middle of being implemented such that patients would naturally be assigned to either an intervention group (PCMH) or control group for post-discharge care. Finally, we employed robust statistical methods to identify and balance confounding variables in the two cohorts through matching. The chief limitations of the study are the modest sample size and the fact that it was a non-randomized study, though we took steps to address biases due to confounding through the matched study design. Nevertheless, not all subjects could be appropriately matched to controls, resulting in a further reduced sample size. We note, for example, that the number of deaths between 6 and 12 months was small, and with a larger sample size, some of the nonsignificant findings here may also have reached statistical significance. For the secondary outcomes of cost and functional questionnaires, we also note that not all subjects responded to the telephone questionnaire or had the appropriate insurance provider that allowed us to measure costs. Finally, the questionnaires were administered at 12 months only and we did not collect preoperative scores, though we expect that acute injury patients would have both EQ-5D and functional scores similar to the general population in their own age range, on average. While we took advantage of the data collection methods of an integrated health system (EHR tracking, claims data, survey center) to perform this investigation, we note that standardized protocols and PCMH models are increasingly being adopted across the U.S. and we believe that these results should be highly translatable to a variety of healthcare settings.

Increasingly, society, patients and payors are demanding measurable quality outcomes. The biggest social impacts may be made by improving the outcome of patient groups whose injuries significantly affect society, such as the elderly patients who sustain hip fractures. Given the multitude of different implant studies which have shown little difference in one-year outcomes for this common yet socially-disruptive injury, studies like this one suggest that the patient care that takes place outside the operating room may ultimately have the biggest impact on improved outcomes. The biggest improvements thus far in hip fracture care have not come from changes in implant technology (e.g., switching to a locked side plate or to an intramedullary device) [30], but from multidisciplinary management of these patients in the acute setting; longer-term benefits of these in-hospital interventions, however, have not been widely demonstrated [32‐34]. It is our hope that the current study provides new evidence regarding important patient outcomes and costs, comparing an intrinsic control group with an experimental group, and that future work will support compelling arguments to government and private payors that the most cost-effective standard of care also gives the highest quality outcome. We recognize that the majority of care-coordination efforts thus far have focused on chronic diseases such as congestive heart failure or diabetes, and that evidence of effectiveness has been mixed, depending on the specific features of the intervention [35‐38]. Future work should determine which specific conditions yield high-value results in order to avoid the error of a one-size-fits-all approach.

Patients receiving aggressive post-discharge care from a “Patient-Centered Medical Home” program showed significant benefits in terms of reduced mortality, improved function, and trends toward reduced medical costs in the time period following hip fracture. Future studies with larger cohorts of PCMH patients are highly encouraged in order to reinforce these results and further examine the details of how PCMH may or may not benefit individual patients. Post-surgical care of elderly, multi-comorbid patients is complex, but these results suggest that ongoing PCMH management can benefit patients in the critical period immediately following a hip fracture and surgery.

A HIPAA waiver of authorization and consent was granted for the retrospective review of records for this study. A partial waiver of authorization and consent was granted for prospectively identifying patients for follow-up questionnaires over the telephone, and completing the telephone survey implied verbal consent on part of the patient.