Introduction
The U.S. Census Bureau projects that by 2030, one in five Americans will be older than 65 years; according to the Centers for Disease Control and Prevention (CDC), one in four will fall each year. Many of these falls result in hip fractures, and 300,000 persons older than 65 years are hospitalized every year with hip fractures (CDC, 2019). Hip fractures result in direct medical costs to the U.S. healthcare system of $5.96 billion (Burge et al., 2006). Fractures in older adults are associated with adverse outcomes such as increased mortality risk, readmissions, institutionalization, chronic pain, and lower health-related quality of life (Alexiou et al., 2018; Downey et al., 2019; Melton, 2003). The complications associated with these hip fractures are the leading cause of injury-related deaths in persons 65 years or older, with a 1-year mortality rate between 14% and 58% (Schnell et al., 2010). In the next two decades, as the older adult population increases, the number of falls and associated costs to care for the injured persons will likely increase in tandem (CDC, 2019). Older adults live longer with more comorbidities; if they sustain a hip fracture, they often have costly and complex medical and pharmacological needs (Linnebur et al., 2005). There is a need to better care for this growing population of patients to help preserve their quality of life and reduce the burden on the healthcare system. A multidisciplinary approach to care is recommended to achieve these desired outcomes (Melton, 2003).
Hip fractures are painful and usually require surgical intervention and pain control for comfort and recovery of physical function. As recommended by the World Health Organization (WHO), analgesics should be prescribed on the basis of pain intensity and adapted to the individual (Anekar & Cascella, 2022; Vargas-Schaffer, 2010). Changes in body composition and organ function are responsible for changes in distribution and clearance of drugs, which can lead to a prolonged plasma elimination half-life. Changes in older adult pharmacodynamics generally tend to increase sensitivity to drugs. Clinicians should be aware of the typical changes in older adult physiology when prescribing pain medications, especially opioids (Anekar & Cascella, 2022; Chau et al., 2008; Linnebur et al., 2005; Mangoni & Jackson, 2004; Vargas-Schaffer, 2010).
Opioids are commonly used for pain management during the acute hip fracture recovery period to provide functional pain management. Functional pain management means patients can participate in rehabilitation therapy and rest comfortably. Typical side effects of opioid medications include nausea, constipation, urinary retention, respiratory depression, pruritus, sedation, and mild cognitive impairment, which can deter patients from participating in therapy and slower their recovery.
Geriatric-Orthopaedic comanagement models are an example of a multidisciplinary approach shown to improve outcomes for patients with fracture and are typically implemented at hospitals where all providers are employees of the medical system. Our academic Level 1 trauma medical center, located in Southern California, cares for more patients older than 80 years than any other academic tertiary healthcare system in California. To better serve our older adult patients, specifically those who sustained a hip fracture, a multidisciplinary team developed their version of a comanagement model called the Geriatric Fracture Program, which was launched in July 2018 (Garlich et al., 2020). The Geriatric Fracture Program was created with the underlying mission to provide optimized, geriatric-centered care that manages the patient's injury in the context of their goals and priorities. A focus of the program was to develop a flexible but standardized pain management protocol appropriate for older adults and provide a multimodal solution offering functional pain control and reducing side effects caused by high morphine milligram equivalent (MME) daily doses. Morphine milligram equivalents were calculated on the basis of the CDC's MME calculator (CDC, 2016). Daily total MMEs are calculated by multiplying the strength of the selected opioid by the total number of milligrams per day with the MME conversion factor (CDC, 2016). Morphine milligram equivalents were used to simplify the comparison between two groups, as the value represents the opioid dose relative to morphine.
To determine whether the intended targets of the Geriatric Pain Protocol (GPP), developed within the larger fracture program, were impacted, we compared patients with hip fracture managed with GPP with those who were not on measures of length of stay (LOS), postoperative complications, and MME.
Methods
Participants and Setting
Our sample (N = 453) included all patients, enrolled in the Geriatric Fracture Program, who were admitted between February 1, 2019, and March 5, 2021. Program inclusion criteria were patients 65 years or older, hip fracture sustained from a ground-level fall, surgical candidate, and care managed by participating physicians. Exclusion criteria included pathological fractures, periprosthetic fractures, nonoperative treatment, and fractures sustained outside of a ground-level fall. The GPP, described in detail later, was a component of the fracture program. Participants treated according to the GPP were compared with those who were not, on LOS, postoperative complications, and MMEs used. Hip fracture types included intertrochanteric, subtrochanteric, and femoral neck fractures treated with an open reduction internal fixation (ORIF), percutaneous pinning, hemiarthroplasty, or total hip arthroplasty. However, the type of fracture or the surgical was not tracked or analyzed. This project was initiated as an operational quality improvement project and was therefore granted an exemption from the institutional review board.
Intervention Protocol
All patients meeting fracture program criteria were treated in accordance with the GPP. The protocol included inclusion in the fracture program, receipt of a preoperative fascia iliaca nerve block, and started on the following pain regimen: acetaminophen 1,000 mg three times daily, celecoxib 200 mg daily postoperatively, and as needed, tramadol 50 mg every 6 hours for moderate pain, oxycodone 5 mg every 4 hours for severe pain, oxycodone 5 mg every 4 hours as needed for breakthrough pain, and hydromorphone 0.4 mg intravenously push every 4 hours as needed for breakthrough pain if not tolerating oral intake. The duration of the fascia iliaca block varies depending on the administration technique and ranges from 8 hours to a median of 48 hours (Levente et al., 2017). The Confusion Assessment Method (CAM) tool was utilized by nursing once each 12-hour shift to assess for postoperative delirium (Inouye et al., 2014; Morrison et al., 2003; Oberhaus et al., 2021). Geriatric Pain Protocol patients had a geriatric specialized nurse practitioner (NP) following care and daily rounds attended by the hospitalist, geriatric NP, primary RN, pharmacist, and case manager. The protocol also included documentation from the electronic medical record by the assigned nurse of all medications given, including type, dosage, route of administration, and time of dose. Pain scores were documented throughout patients' stays and were documented upon admission, Postoperative Day 1 (POD1), and POD2. Figure 1 shows the pharmacological pain protocol, and Figure 2 shows nonpharmacological pain management measures.
Variables
Patient demographics were constrained to age and biological sex to reduce the number of variables. Following is a comprehensive list of variables, including the assessment tool used (where applicable) and measurement: delirium assessment using the CAM, medications, pain using the Numerical Rating Scale (NRS), nonpharmacological pain management, inpatient complications, 30-day readmissions, and time to care intervals. A detailed explanation of each variable is as follows.
Delirium
Assessment using the CAM was performed by an MD or nurse and scored as positive or negative (for the presence of acute change of mentation, inattention, and disorganized thinking and/or change in alertness) on each patient upon admission (preoperatively) and again upon returning from the operating room (OR) (postoperatively) (Inouye et al., 2014). This was continued by the nurse assigned to the patient each 12-hour shift thereafter.
Medications
Daily totals for the MMEs were calculated for POD1 beginning at midnight POD1 through midnight of POD2.
Pain
Pain score was measured using the NRS, a self-reported measure of pain intensity ranging from 0 to 10 (Herr & Garand, 2001; Karcioglu et al., 2018).
Nonpharmacological Pain Management
Nonpharmacological pain management strategies, such as ice, distraction, and repositioning, were offered by the nurse and on the comfort menu available for patients to request from their nurse (see Figure 2 for the menu provided to the patients).
Inpatient Complications
Complications were chosen on the basis of extant literature, clinical experience, and expertise and identified postoperatively via chart review (Carpintero et al., 2014; Groff et al., 2020; Kastanis et al., 2016; Nguyen & Lui, 2020; Schnell et al., 2010). Data on issues or concerns related to the complications listed in the following were collected and counted. On the basis of the data distribution and for ease of interpretation, the variable "inpatient complication" was dichotomized into 0 versus >=1.
* Psychological status
* Renal function
* Acute kidney injury, chronic kidney disease, urinary tract infections, sepsis due to urinary tract infection, urinary retention
* Transfusion
* Underlying acute blood loss or anemia
* Respiratory distress
* Aspiration pneumonia, acute respiratory failure, required postoperative intubation, BiPAP (bilevel positive airway pressure)
* Deep vein thrombosis/pulmonary embolism
* Electrolyte disturbances
* Hypoglycemia, hyponatremia, hypo/hyperkalemia
* Gastrointestinal health
* Constipation, low oral intake, nausea/vomiting, dysphasia, gastric tube placement, aspiration, gastrointestinal bleed, perforated bowl, ileus, rectal spasms
* Infection/bacteremia
* Cardiac
* Congestive heart failure exacerbation, arrhythmias
* Neurological function
* Hallucinations, altered mental status/encephalopathy, agitation, stroke, lethargy, Narcan use
* Uncontrolled pain
* Reblocked postoperatively, need for epidural, peripheral nerve line use
* Bleeding
* Hematoma, retroperitoneal hemorrhage, thrombocytopenia
* Death
30-Day Readmission
The number of patients readmitted to our hospital within 30 days from discharge was compared between the GPP and non-GPP groups.
Timing of Care Intervals
Intervals calculated were inpatient admission into the OR, nerve block procedure to the OR, OR to first inpatient physical therapy (PT) session, and inpatient admission to discharge from the hospital, which is the LOS (American Academy of Orthopaedic Surgeons, 2021).
End Points, Study Aims, and Statistical Analysis
The primary aim of this study was to compare patients whose care followed the GPP guidelines versus those whose care did not follow the GPP guidelines with respect to clinically and procedurally relevant end points. End points included, but were not limited to, postoperative pain, postoperative opioid utilization, inpatient delirium, LOS, complications, and 30-day readmission rates.
As true population parameters were unknown, nonparametric statistical tools were used to compare groups (Hopkins et al., 2018). The Mann-Whitney U test was used to compare continuous and ordinal variables between groups. Fisher's exact test was used to compare binary or categorical covariates between groups. Spearman's rank-based correlation coefficient was used to assess associations between two continuous/ordinal variables.
Delirium was assessed preoperatively and postoperatively using the CAM as detailed in the "Variables" section. A secondary analysis utilized multiple logistic regression to compare the GPP and non-GPP groups in terms of postoperative delirium while adjusting for preoperative delirium status. All statistical analyses were performed using the statistical computing language R, Version 4.0.5 (R Core Team, Vienna, Austria; access date: July 18, 2021).
Statistical power was considered for each type of planned hypothesis test, given the known sample size. Two-tailed nonparametric hypothesis testing and an [alpha] level of .05 were assumed; 220 patients per group were sufficient to detect an effect size of Cohen's d = 0.274 when performing a Mann-Whitney U test to compare a continuous or ordinal variable between groups, and an effect size of Cohen's w = 0.134 when testing a 2 x 2 contingency table. Both effect sizes are conventionally considered to be small-medium (Leydesdorff et al., 2019). This study cannot rule out subtler between-group effects than these respective effect sizes in the case of nonstatistically significant hypothesis tests.
Results
Group Comparisons
Table 1 shows details of group demographics by GPP participation and overall. The sample was 69% female (n = 311), and the mean age was 84 +/- 8.8 years; provider specialty and age were the only between-group differences. Preoperative delirium was present in 6.2% (n = 28) of the sample.
Table 2 shows group comparisons on outcomes of interest. Opioid usage in MMEs was lower in the GPP group than in the non-GPP group for both POD1 (p = .007) and POD2 (p = .043); NRS Pain on POD1 was also lower in the GPP group (vs. non-GPP, p = .013). There were no between-group differences in the occurrence of postoperative delirium (CAM), NRS Pain on POD2, inpatient complications, or 30-day readmissions (all ps > .1; see Table 2).
There were no differences between groups with respect to any time interval variables, including LOS (all ps > .3; results not shown). The breakdown of medical providers significantly differed between the GPP and non-GPP patient groups (Fisher's exact test, p = .011). The medical center employs private hospitalist groups, independent community physicians, and a group of salaried faculty hospitalists accounting for the significant difference noted. Patients in the GPP group had a higher median age (87 years) than the non-GPP group (84 years; Mann-Whitney U test, p = .041). There was no difference in the proportion of males and females depending on the group (GPP or non-GPP; Fisher's exact test, p = .48). There was no difference in the prevalence of preoperative delirium between the GPP and non-GPP groups (Fisher's exact test, p = .21).
Regression Analysis
A factorial interaction term between the group and baseline CAM score was tested against a main-effects model using the chi-squared test. It was found that the interaction term did not add significantly to the model, illustrating that the effect of baseline CAM score is independent of the GPP group, and vice versa. Unsurprisingly, in the main-effects model, the preoperative CAM score was strongly predictive of the postoperative CAM score (p < .001). When holding the baseline CAM score constant, there was no evidence that the GPP was protective against postoperative CAM (p = .83).
Correlation Analysis
Pain and Postoperative MME
MME and NRS Pain levels on POD1 and POD2 were correlated with higher age (rho [[rho]] values -0.19 to -0.32, all ps < .001), and a higher NRS Pain score was also correlated with higher MME.
Time Intervals
Increased LOS was subtly correlated with higher age ([rho]= 0.10, p = .029), increased time from admission to nerve block ([rho]= 0.19, p < .001), and from OR to PT ([rho]= 0.29, p < .001). However, increased time from admission to nerve block was strongly correlated with decreased time from nerve block to OR ([rho]= -0.58, p < .001); we do not consider it relevant clinically, as the association is outside the scope of the GPP and could be due to various workflows, staffing, and patient factors. A few examples include the following: the order being forgotten and not placed, the regional anesthesia pain team being behind schedule or overburdened, or the patient initially refusing and then later requesting or authorizing the nerve block.
Discussion
As the U.S. population ages, more falls resulting in hip fractures will occur, mainly because falls that occur at an advanced age are more likely to result in fractures (Ensrud, 2013; Kang et al., 2010). Hip fractures are painful and usually require surgical intervention and pain control for comfort and recovery of physical function. As recommended by the WHO, analgesics should be prescribed on the basis of pain intensity and adapted to the individual while considering older adult physiological changes and their effects on pharmacokinetics and pharmacodynamics (Chau et al., 2008; Vargas-Schaffer, 2010).
Most patients with hip fracture do not return to their previous level of health or mobility. Still, by providing functional pain control and working with patients to determine their best-case scenario for recovery, they can return more quickly to an acceptable quality of life (Alexiou et al., 2018). One way to manage pain that considers the older adults' physiology and mitigates medication side effects and complications from the treatment is to develop a standardized but flexible multimodal pain management practice. We assessed the impact of the GPP, a multimodal pain management protocol designed to provide functional pain management, and found the GPP resulted in lower MME and equivalent pain management for our patients with hip fractures.
Although median age differed between groups, the ranges were comparable, and studies show no differences in healing rates and time to union between those younger than 65 years and those who are older than 65 years (Taormina et al., 2014). Thus, we considered this finding of minimal clinical importance when interpreting the remainder of our findings.
That there was higher reported pain among non-GPP patients on POD1, which may have been due to the residual pain reduction from the nerve block received by the GPP group, comports with earlier findings from our institution (Garlich et al., 2020). The higher pain score could partially explain the higher MME for the non-GPP group on POD1. Still, both groups had equivalent reported pain on POD2, and the GPP group again showed lower MME, which calls into question the relationship of the higher pain score with a requirement for the use of higher MME medications on POD1.
Although our findings are important, some limitations bear mentioning. First, data were collected in only one inpatient unit, though all patients meeting inclusion criteria were included, improving the likelihood of the representativeness of our sample to the underlying population of patients. Second, this was a quality improvement initiative and was therefore not powered to detect a difference, nor were the participants randomized. However, the GPP was available to all patients and whether a patient was treated in accordance with the GPP was due to circumstances not related to the protocol and as evidenced by the resulting distribution of patients between groups, resembles random assignment, ameliorating concerns about the potential for Type I error. In addition, anesthetic techniques and analgesics administered intraoperatively, which can potentially impact postoperative pain and delirium, were not considered. Finally, lending confidence to our findings is the fact that the likelihood of finding a difference is diminished when a study is not powered to do so, and reductions in MME were still apparent.
This multimodal pain protocol for geriatric patients who sustained a fracture from a ground-level fall was shown to result in lower MME. This is important for patients 65 years or older for many reasons, not the least of which is that hip fractures are strongly associated with mortality and diminished quality of life (Alexiou et al., 2018). Older hospitalized adults are at risk of deconditioning and functional decline (Kresevic, 2015). Regular physical activity during hospitalization is vital to help patients regain their functional abilities. Without adequate pain control, patients either do not want to or cannot mobilize and participate in rehabilitation therapy (Caba et al., 2022).
With the aging U.S. population, it is imperative to find ways to better treat these patients so they can return to their acceptable quality of life. Doing so will improve outcomes not only for patients but also for the healthcare system with reduced costs and overutilization. Furthermore, in the age of the Bundled Payments for Care Improvement Advanced (BPCI Advanced) from the Centers for Medicare & Medicaid Services, it is fiscally important for healthcare organizations to improve the quality and value of care to reduce LOS and the frequency of readmissions. Quality care through protocols such as these and multidisciplinary approaches to patient care can help achieve just that.
References