Abstract
BACKGROUND:
It is not clear if there is a risk of 30-day readmissions following total hip and knee arthroplasty in patients reporting high levels of pain at hospital discharge. We examined the relationship between post-surgical pain on the day of discharge and 30-day readmission in patients who received total knee and hip arthroplasty.
METHODS:
Retrospective cohort study of patients who received total knee (n = 155,284) or hip arthroplasty (n = 89,283) from 2011 to 2018 using electronic health records (EHR) from the Optum database. Four categories of pain at discharge were created, from none to severe. Multivariate logistic regression models to predict 30-day all cause readmission were adjusted for patient and clinical characteristics and built separately for knee and hip arthroplasty patients.
RESULTS:
Mean ages for hip and knee patients were 64.4 (SD, 11.3) and 65.7 (SD, 9.7) years, respectively. A majority were female (hip: 54.4%, knee: 61.5%). The unadjusted rate of 30-day readmission was 3.54% for hip replacement and 3.66% for knee replacement. In models adjusted for patient and clinical characteristics, for patients with total hip replacement, the odds of 30-day readmission for those with severe pain score at discharge versus those with no pain at discharge were 1.60 (95% confidence interval [CI] 1.33–1.92). Similarly, readmission likelihood increased as pain at discharge increased (severe pain versus no pain) for patients with total knee arthroplasty (OR 1.38, 95% CI 1.19–1.59).
CONCLUSION:
Our findings demonstrated that the pain scores on the day of discharge are associated with 30-day hospital readmission.
Keywords: pain, hip arthroplasty, knee arthroplasty, 30-day readmission, electronic health records
Introduction
Hospital readmission impairs patient health outcomes and increases medical costs [1]. In the United States, 30-day readmissions cost about $17 billion annually [2] and two-thirds of these hospital readmissions can be prevented [3]. Under the Center for Medicare and Medicaid Services (CMS) Value Based Purchasing Program, readmission rates are used as an important measure of quality across diagnostic conditions and hospitals [4]. Patients receiving elective lower extremity joint replacement are one of six diagnostic groups identified by CMS’s Hospital Readmissions Reduction Program [5]. Hospitals that fail to meet readmission benchmarks are subject to financial penalties. Another example is the Comprehensive Care for Joint Replacement (CJR) program introduced by the CMS to improve care transitions after lower extremity joint replacement [6].
Previous studies using administrative claims data report that various patient, clinical and facility factors such as age, race/ethnicity, gender, socioeconomic status, discharge disposition, type of facility and geographic location are associated with higher readmission rates [7–21]. A recent study involving Medicare beneficiaries receiving elective lower extremity joint replacements reported that commonly used comorbidity indices (e.g., Elixhauser) included in CMS risk adjustment models were not significantly associated with discharge destination or the risk of hospital readmission [22]. These findings suggest the need for future research including the study of measures such as obesity, functional status or pain related to readmission rates to improve the CMS Hospital Readmission Reduction Program [23].
Previous studies reported that patients with pain-related medical conditions have higher a rate of readmission (i.e., those with arthritis who received total hip arthroplasty or total knee arthroplasty) [8, 12, 16, 20]. However, very limited information is available on the association of post-surgery pain measures and 30-day readmission following total hip arthroplasty or total knee arthroplasty.
Many large studies of hospital readmissions have used administrative claims data (e.g., Medicare claims) [9–11, 14, 16, 17, 20]. Medicare data associated with acute hospitalizations do not have easily available information about post-surgical measures of pain [9–11]. The purpose of Medicare acute care hospital files is to determine billing. Patient-level electronic health records (EHR) contain more in-depth patient-level healthcare information (e.g., pain, social behaviors). Patient-level data on pain can be extracted from physician notes, lab results or measurement and assessment values in the EHR. These records, in the form of research data, are now available for research purposes from data warehouses and repositories, but have not been widely used to study the association of pain and 30-day readmission in diagnostic groups related to the Hospital Readmissions Reduction Program. The purpose of this study was to examine the association of pain on the day of discharge and 30-day readmission related to pain-related medical conditions in patients who received a total hip arthroplasty or total knee arthroplasty. We hypothesized that post-surgical measures of pain available in the EHR and obtained on the day of discharge would be significantly associated with 30-day readmission. We also hypothesized that discharge destinations would be an important mediating or moderating variable in the relationship between pain and 30-day readmission.
Methods
Data Sources
The study used de-identified data from Optum’s EHR database. The Optum EHR data includes more than 90 million patients from 38 hospital networks and 18 non-network hospitals in the United States. The database contains structured information, such as diagnosis, procedure codes, lab results and observations, including vital signs, blood pressure, pain and body mass index (BMI). It also contains unstructured information in the form of clinical notes from office visits, consultation reports, discharge summaries and reports from nursing records, pathology, radiology and cardiology. Using a natural language processing (NLP) system, the unstructured data in the clinical notes are scanned to discover, interpret and extract important clinical information. This information then gets organized into structured fields for clinical assessment in tables of signs/symptoms, treatment rationale and clinical measurements [24, 25]. Clinical measurements include numeric fields from clinical notes such as blood pressure, pain and BMI values. A data use agreement was developed with Optum and the study was approved by the local Institutional Review Board.
Patient Cohort
The study sample is based on a cohort from Optum’s EHR data consisting of patients who underwent lower extremity bone or joint surgery between January 1, 2011 and February 28, 2018. The study sample was discharged from the hospital on or before February 28, 2018. Figure 1 presents the cohort section process. From this cohort of patients, we selected those who had a total hip arthroplasty or total knee arthroplasty in inpatient settings, using International Classification of Diseases, Ninth and 10th Revision (ICD-9 and ICD-10) codes (ICD-9-PCS and ICD-10-PCS) and Current Procedure Terminology (CPT) codes (Table A.1). Multiple inpatient encounters comprising a single hospital stay were aggregated into an inpatient visit by Optum, creating information on admission and discharge dates, which were not typically included in the original EHR data.
Fig. 1.
Cohort selection.
We excluded surgeries that did not have complete admission or discharge data and those classified as revision surgeries. From the surgeries studied, we selected the first surgery for each patient (either hip or knee) and excluded those without complete data. The final cohort included 89,283 patients with a total hip arthroplasty and 155,284 patients with a total knee arthroplasty.
Outcome
The primary outcome was 30-day all-cause readmission (dichotomous, yes/no) from the index hip or knee total joint arthroplasty surgery, as determined from the inpatient admission record [26].
Primary Predictors
Our primary variables of interest were patient-reported pain level on the day of hospital discharge as measured by a visual analogue scale (0 – No pain to 10 – Unbearable pain). Pain data in the Optum EHR are taken from clinical observations. We created four categories based on the pain score distribution (no pain, 1–3, 4–6, 7–10). The pain variable at discharge from the 0-to-10 visual analogue scale was use when available (97.3% of pain values). When the 0-to-10 structured data was not available, pain was obtained from the NLP-created structured fields for clinical measurements. This was done by selecting measurement records where the assessment term was “pain.” Values were excluded that were assessed on a different scale (e.g., 0 to 100) or were not numeric (e.g., “increased”).
Covariates
Patient characteristics included in the analysis were gender (male, female), race (Caucasian, African American, Asian, other), ethnicity (Hispanic, not Hispanic, unknown) and age (<55, 55–64, 65–74, 75+ years). For hospital length of stay and Charlson Comorbidity Index [27], we examined the distributions to create a categorical variable for each. We created a four-category variable for hospital length of stay (0, 1, 2, 3, 4+ days) and a three-category variable for the Charlson Comorbidity Index (0, 1, 2, 3+), by collapsing the right tail of distributions to have the highest level with proportion between 15–20%. We also grouped the patients into four diagnosis related groups (DRGs) (470: primary arthroplasty without major complication/comorbidity [MCC]; 462: bilateral or multiple primary arthroplasty without MCC; 469: primary arthroplasty with MCC; Other: including all other diagnoses) to examine the association of pain discharge with cause-specific readmission [28]. BMI was obtained similar to pain, in that it was assessed first from the structured data and then from the NLP-created structured fields for clinical measurements when not available in structured fields [24, 25]. Numeric BMI was classified into four categories (underweight: <18.5; normal: 18.5–24.9; overweight: 25.0–29.9; obese: ≥30.0) based on the Centers for Disease Control and Prevention guideline [29]. BMI values were assessed between 30 days prior to surgery and hospital discharge, with the value closest to the day of surgery being used. Also, we provided descriptive information about condition category variables (yes, no) used in the CMS as risk adjustors for calculating the Risk-Standardized Readmission Rate (RSRR). These include 27 conditions such as arrhythmia, atherosclerosis and cancer (Table 1) [30]. However, instead of individual condition categories, the Charlson Comorbidity Index scores were used as a risk adjustor in the multivariable data analyses. These condition category variables were assessed in the 12 months prior to the joint arthroplasty. Community discharge was defined as a discharge to the community (to home or with home health care).
Table 1.
Patient Characteristics
| Hip | Knee | |||
|---|---|---|---|---|
| N | % | N | % | |
| All | 89,283 | 100 | 155,284 | 100 |
| Gender | ||||
| Female | 48,568 | 54.4 | 128,517 | 61.5 |
| Male | 40,715 | 45.6 | 80,621 | 38.6 |
| Race | ||||
| Caucasian | 79,875 | 89.5 | 136,749 | 88.1 |
| African American | 5,975 | 6.7 | 10,907 | 7.0 |
| Asian | 349 | 0.4 | 1,036 | 0.7 |
| Other/Unknown | 3,084 | 3.5 | 6,592 | 4.3 |
| Ethnicity | ||||
| Hispanic | 1,730 | 1.9 | 4,302 | 2.8 |
| Not Hispanic | 83,993 | 94.1 | 145,410 | 93.6 |
| Unknown | 3,560 | 4.0 | 5,572 | 3.6 |
| Age (years), mean SD | 64.4 | 11.3 | 65.7 | 9.7 |
| Age (years) | ||||
| <55 | 16,081 | 18.0 | 19,296 | 12.4 |
| 55–64 | 28,236 | 31.6 | 50,360 | 32.4 |
| 65–74 | 27,391 | 30.7 | 55,147 | 35.5 |
| 75+ | 17,575 | 19.7 | 30,481 | 19.6 |
| Pain | ||||
| None | 8,683 | 9.7 | 9,281 | 6.0 |
| 1–3 | 48,350 | 54.2 | 73,006 | 47.0 |
| 4–6 | 28,491 | 31.9 | 63,569 | 40.9 |
| 7–10 | 3,759 | 4.2 | 9,428 | 6.1 |
| Body Mass Index | ||||
| Underweight | 713 | 0.8 | 321 | 0.2 |
| Normal | 16,865 | 18.9 | 13,915 | 9.0 |
| Overweight | 29,326 | 32.9 | 41,016 | 26.4 |
| Obese | 42,379 | 47.5 | 100,032 | 64.4 |
| Community Discharge | 73,462 | 82.3 | 125,012 | 80.5 |
| Charlson Comorbidity Index | ||||
| 0 | 43,207 | 48.4 | 68,591 | 44.2 |
| 1 | 18,775 | 21.0 | 37,800 | 24.3 |
| 2 | 12,192 | 13.7 | 21,181 | 13.6 |
| 3+ | 15,109 | 16.9 | 27,712 | 17.9 |
| Hospital Length of Stay (Days) | ||||
| 1 | 19,350 | 21.7 | 18,255 | 11.8 |
| 2 | 34,718 | 38.9 | 63,413 | 40.8 |
| 3 | 25,318 | 28.4 | 54,672 | 35.2 |
| 4+ | 9,897 | 11.1 | 18,944 | 12.2 |
| Year of Surgery | ||||
| 2011 | 392 | 0.4 | 766 | 0.5 |
| 2012 | 8,044 | 9.0 | 15,971 | 10.3 |
| 2013 | 11,599 | 13.0 | 22,079 | 14.2 |
| 2014 | 14,101 | 15.8 | 25,606 | 16.5 |
| 2015 | 16,462 | 18.4 | 28,615 | 18.4 |
| 2016 | 18,220 | 20.4 | 29,770 | 19.2 |
| 2017 | 17,898 | 20.1 | 28,765 | 18.5 |
| 2018 | 2,567 | 2.9 | 3,712 | 2.4 |
| Diagnosis Related Group (DRG) | ||||
| 470 | 77,242 | 86.5 | 129,758 | 83.6 |
| 462 | 565 | 0.6 | 6,773 | 4.4 |
| 469 | 1,503 | 1.7 | 2,303 | 1.5 |
| Other | 1,172 | 1.3 | 1,904 | 1.2 |
| Missing | 8,801 | 9.9 | 14,546 | 9.4 |
| Number of Condition Categories, mean (SD) | 2.72 | 2.3 | 3.08 | 2.3 |
| Condition Categories | ||||
| Arrhythmia | 14,003 | 15.7 | 26,498 | 17.1 |
| Atherosclerosis | 11,956 | 13.4 | 22,154 | 14.3 |
| Cancer | 8,963 | 10.0 | 15,347 | 9.9 |
| Cellulitis | 2,444 | 2.7 | 4,818 | 3.1 |
| Congestive Heart Failure | 3,574 | 4.0 | 6,559 | 4.2 |
| Chronic Obstructive Pulmonary | ||||
| Disease | 7,919 | 8.9 | 13,034 | 8.4 |
| Dementia | 1,719 | 1.9 | 2,910 | 1.9 |
| Diabetes | 14,692 | 16.5 | 35,925 | 23.1 |
| Dialysis | 110 | 0.1 | 155 | 0.1 |
| Fluid or Metabolic Disorder | 7,775 | 8.7 | 13,033 | 8.4 |
| Hemolytic Disorder | 241 | 0.3 | 298 | 0.2 |
| Hypertension | 53,792 | 60.3 | 106,680 | 68.7 |
| Major Symptoms, Abnormalities | 46,377 | 51.9 | 87,909 | 56.6 |
| Malnutrition | 535 | 0.6 | 446 | 0.3 |
| Metastatic Cancer | 484 | 0.5 | 413 | 0.3 |
| Morbid Obesity | 9,021 | 10.1 | 25,997 | 16.7 |
| Other Injuries | 12,265 | 13.7 | 28,375 | 18.3 |
| Paralysis | 872 | 1.0 | 1,314 | 0.9 |
| Pneumonia | 1,538 | 1.7 | 2,656 | 1.7 |
| Polyneuropathy | 8,433 | 9.5 | 16,999 | 11.0 |
| Psychiatric | 4,269 | 4.8 | 8,341 | 5.4 |
| Renal Failure | 7,139 | 8.0 | 13,363 | 8.6 |
| Rheumatoid Arthritis | 6,189 | 6.9 | 10,608 | 6.8 |
| Severe Infection | 7,304 | 8.2 | 14,008 | 9.0 |
| Stroke | 994 | 1.1 | 1,711 | 1.1 |
| Ulcer | 781 | 0.9 | 1,268 | 0.8 |
| Vascular Disorder | 9,608 | 10.8 | 16,976 | 10.9 |
Note. DRG 470, major hip and knee joint replacement or reattachment of lower extremity w/o major complication or comorbidity; DRG 462, bilateral or multiple major joint procedures of lower extremity without major complication or comorbidity; DRG 469, major hip and knee joint replacement or reattachment of lower extremity w mcc or total ankle replace
Analysis
We used multivariate logistic regression models to assess the association of pain scores on the day of hospital discharge and hospital readmission within 30 days of hip or knee arthroplasty surgery. We assessed the association using an unadjusted model and an adjusted model controlling for patient characteristics, length of stay, Charlson Comorbidity score categories, surgery year and community discharge. We conducted a sensitivity analysis that included only patients with DRG available. The sensitivity analysis was done separately for hip and knee arthroplasty. Statistical significance was set at p < 0.05. All analyses were performed using SAS statistical software version 9.4 (SAS Institute., Cary, NC).
Results
Cohort Characteristics
Table 1 presents the characteristics of the 89,283 patients with total hip arthroplasty surgery and the 155,284 with total knee arthroplasty surgery. The average age of the total hip and knee arthroplasty patients was 64.4 (standard deviation [SD], 11.3) and 65.7 (SD, 9.7) years, respectively. A majority were female (hip: 54.4%, knee: 61.5%), Caucasian (hip: 89.5%, knee: 88.1%), not Hispanic (hip: 94.1%, knee: 93.6%), obese (hip: 47.5%, knee: 64.4%) and discharged to the community from the hospital (hip: 82.3%, knee: 80.5%).
Readmissions
In the unadjusted models, the overall rate of 30-day readmission was 3.54% for total hip arthroplasty and 3.66% for total knee arthroplasty. Tables 2 and 3 present the odds ratios (OR) of 30-day readmission after total hip or knee arthroplasty surgery based on pain on hospital discharge. In the unadjusted models (Model I), the odds of 30-day readmission increased with a high pain score of 7–10 compared to no pain in both total hip arthroplasty (OR 1.85, 95% confidence interval [CI] 1.55–2.21) and total knee arthroplasty (OR 1.31, 95% CI 1.14–1.52).
Table 2.
Association between post-surgery pain level hospital readmission within 30 days of hip replacement surgery, by joint of surgery.
| Effect | Level | Readmission % | Model I Unadjusted OR (95% CI) |
Model II Adjusted OR (95% CI) |
|---|---|---|---|---|
| Pain | None | 3.40 | Ref. | Ref. |
| 1–3 | 3.02 | 0.89 (0.78, 1.01) | 0.99 (0.87, 1.12) | |
| 4–6 | 4.13 | 1.23 (1.08, 1.40) | 1.24 (1.08, 1.41) | |
| 7–10 | 6.12 | 1.85 (1.55, 2.21) | 1.60 (1.33, 1.92) | |
| Age | <55 | 3.15 | Ref. | |
| 55–64 | 3.07 | 0.93 (0.83, 1.05) | ||
| 65–74 | 3.14 | 0.87 (0.77, 0.97) | ||
| 75+ | 5.29 | 1.16 (1.03, 1.32) | ||
| Gender | Male | 3.32 | Ref. | |
| Female | 3.73 | 1.01 (0.93, 1.08) | ||
| Race | Caucasian | 3.57 | Ref. | |
| African American | 3.80 | 0.89 (0.78, 1.03) | ||
| Asian | 3.15 | 0.92 (0.50, 1.69) | ||
| Other | 2.37 | 0.70 (0.54, 0.90) | ||
| Ethnicity | Non-Hispanic | 3.59 | Ref. | |
| Hispanic | 2.83 | 0.91 (0.67, 1.23) | ||
| Unknown | 2.72 | 0.86 (0.70, 1.07) | ||
| BMI | Normal | 4.91 | REF | |
| Underweight | 3.28 | 1.29 (0.90, 1.83) | ||
| Overweight | 3.16 | 1.02 (0.91, 1.13) | ||
| Obese | 3.88 | 1.21 (1.09, 1.34) | ||
| Charlson Comobidity | 0 | 2.51 | Ref. | |
| 1 | 3.53 | 1.25 (1.14, 1.39) | ||
| 2 | 3.95 | 1.39 (1.24, 1.55) | ||
| 3+ | 6.19 | 1.95 (1.77, 2.14) | ||
| Length of Stay | 1 | 2.23 | Ref. | |
| 2 | 2.97 | 1.23 (1.09, 1.38) | ||
| 3 | 3.99 | 1.31 (1.15, 1.49) | ||
| 4+ | 6.98 | 1.99 (1.73, 2.29) | ||
| Surgery Year | 2011 | 6.63 | Ref. | |
| 2012 | 3.84 | 0.63 (0.41, 0.96) | ||
| 2013 | 3.57 | 0.60 (0.40, 0.91) | ||
| 2014 | 3.58 | 0.64 (0.42, 0.96) | ||
| 2015 | 3.51 | 0.63 (0.42, 0.95) | ||
| 2016 | 3.59 | 0.65 (0.43, 0.98) | ||
| 2017 | 3.34 | 0.62 (0.41, 0.94) | ||
| 2018 | 3.08 | 0.58 (0.37, 0.93) | ||
| Community Discharge | No | 6.60 | Ref. | |
| Yes | 2.88 | 0.63 (0.57, 0.69) |
Note. Model II included pain, age, gender, race, ethnicity, body mass index, Charlson Comorbidity score, length of hospital stay, surgery year, and community discharge.
Table 3.
Association between post-surgery pain level hospital readmission within 30 days of knee replacement surgery, by joint of surgery.
| Effect | Level | Readmission % | Model I Unadjusted OR (95% CI) |
Model IIa Adjusteda OR (95% CI) |
|---|---|---|---|---|
| Pain | None | 3.75 | Ref. | Ref. |
| 1–3 | 3.30 | 0.88 (0.78, 0.98) | 0.98 (0.88, 1.10) | |
| 4–6 | 3.88 | 1.04 (0.92, 1.16) | 1.16 (1.04, 1.31) | |
| 7–10 | 4.87 | 1.31 (1.14, 1.52) | 1.38 (1.19, 1.59) | |
| Age | <55 | 3.39 | Ref. | |
| 55–64 | 3.14 | 0.92 (0.83, 1.01) | ||
| 65–74 | 3.48 | 0.98 (0.89, 1.07) | ||
| 75+ | 5.01 | 1.27 (1.15, 1.41) | ||
| Gender | Male | 4.19 | Ref. | |
| Female | 3.33 | 0.74 (0.70, 0.79) | ||
| Race | Caucasian | 3.65 | Ref. | |
| African American | 4.51 | 1.16 (1.05, 1.28) | ||
| Asian | 2.80 | 0.74 (0.51, 1.08) | ||
| Other | 2.62 | 0.72 (0.61, 0.86) | ||
| Ethnicity | Non-Hispanic | 3.69 | Ref. | |
| Hispanic | 3.42 | 1.05 (0.88, 1.25) | ||
| Unknown | 2.93 | 0.88 (0.75, 1.04) | ||
| BMI | Normal | 4.98 | Ref. | |
| Underweight | 3.90 | 1.24 (0.74, 2.07) | ||
| Overweight | 3.64 | 0.93 (0.84, 1.03) | ||
| Obese | 3.63 | 0.94 (0.85, 1.03) | ||
| Charlson Comorbidity | 0 | 2.79 | Ref. | |
| 1 | 3.38 | 1.17 (1.09, 1.26) | ||
| 2 | 4.08 | 1.38 (1.27, 1.50) | ||
| 3+ | 5.86 | 1.89 (1.76, 2.03) | ||
| Length of Stay | 1 | 2.71 | Ref. | |
| 2 | 3.33 | 1.15 (1.04, 1.27) | ||
| 3 | 3.64 | 1.13 (1.02, 1.26) | ||
| 4+ | 5.73 | 1.61 (1.44, 1.82) | ||
| Surgery Year | 2011 | 2.87 | Ref. | |
| 2012 | 3.68 | 1.35 (0.88, 2.08) | ||
| 2013 | 3.80 | 1.41 (0.92, 2.17) | ||
| 2014 | 3.67 | 1.37 (0.89, 2.11) | ||
| 2015 | 4.02 | 1.51 (0.98, 2.32) | ||
| 2016 | 3.88 | 1.46 (0.95, 2.24) | ||
| 2017 | 3.05 | 1.14 (0.74, 1.75) | ||
| 2018 | 2.96 | 1.10 (0.69, 1.76) | ||
| Community Discharge | No | 5.21 | Ref. | |
| Yes | 3.28 | 0.77 (0.72, 0.82) |
Note. Model II included pain, age, gender, race, ethnicity, body mass index, Charlson Comorbidity score, length of hospital stay, surgery year, and community discharge.
After adjusting for patient characteristics (Model II), the OR of 30-day readmission for patients with total hip arthroplasty and a pain score of 7–10 versus those with no pain was 1.60 (95% CI 1.33–1.92). Similarly, the odds of 30-day readmission also increased as pain on discharge increased for patients with total knee arthroplasty, after adjusting for patient characteristics (OR 1.38, 95% CI 1.19–1.59).
Females were less likely to be readmitted within 30 days post-surgery among patients with total knee arthroplasty (Model II, OR 0.74, 95% CI 0.70–0.79); however, this relationship was not significant for patients with total hip arthroplasty (Model II, OR 1.01, 95% CI 0.93–1.08). Odds of 30-day readmission increased in adults over 75 years versus adults younger than 55 years for both total hip arthroplasty (Model II, OR 1.16, 95% CI 1.03–1.32) and total knee arthroplasty (Model II, OR 1.27, 95% CI 1.15–1.41). Odds of 30-day readmission decreased in other races compared to Caucasians for both total hip arthroplasty (Model II, OR 0.70, 95% CI 0.54–0.90) and total knee arthroplasty ((Model II, OR 0.72, 95% CI 0.61–0.86), but odds of 30-day readmission increased in African Americans compared to Caucasians for total knee arthroplasty (Model II, OR 1.16, 95% CI 1.05–1.28).
Odds of 30-day readmission increased for patients with Charlson comorbidity score greater than 3 versus those with Charlson comorbidity score 0 for both total hip arthroplasty (Model II, OR 1.95, 95% CI 1.77–2.14) and total knee arthroplasty (Model II, OR 1.89, 95% CI 1.76–2.03). Similarly, odds of 30-day readmission increased for patients with length of stay greater than 4 days versus those with a 1-day length of stay for both total hip arthroplasty (Model II, OR 1.99, 95% CI 1.73–2.29) and total knee arthroplasty (Model II, OR 1.61, 95% CI 1.44–1.82).
Among patients with total hip arthroplasty, the odds of 30-day readmission for patients who were obese versus those with normal weight were 1.21 (95% CI 1.09–1.34), after adjusting for patient characteristics (Model II). However, the odds were non-significant (OR 0.94, 95% CI 0.85–1.03) among patients with total knee arthroplasty (Model II). Lastly, odds of 30-day readmission decreased in patients discharged to the community for both total hip arthroplasty (Model II, OR 0.63, 95% CI 0.57–0.69) and total knee arthroplasty (Model II, OR 0.77, 95% CI 0.72–0.82).
We conducted a sensitivity analysis to examine the association of the pain measures on the day of discharge with cause-specific readmission using the DRG groups. Odds of 30-day readmission increased for patients with DRG 469 (total joint arthroplasty with comorbidity) versus those with DRG 470 (total joint arthroplasty without comorbidity) for both total hip arthroplasty (Model II, OR 1.50, 95% CI 1.24–1.81) and total knee arthroplasty (Model II, OR 1.50, 95% CI 1.28–1.76) (Tables A.2 and A.3).
Discussion
The overall rate of 30-day readmission in our study was 3.54% for total hip arthroplasty and 3.66% for total knee arthroplasty in the unadjusted models. These rates are lower than the 6.9% for total hip arthroplasty and 5.9% for total knee arthroplasty previously reported in Medicare beneficiaries [30]. The difference in age ranges between the Optum EHR participants in our sample (mean hip 64.5 years, mean knee 65.7 years) and the greater age of Medicare beneficiaries (mean 75.1 years) is likely a factor in the difference in readmission rates. Similarly, in our study, the 30-day readmission in patients with total knee arthroplasty increased between 2014 and 2016. However, a study using patients 50 years or older in the Nationwide Readmissions Database reported that the 30-day readmission rates decreased between 2010 and 2015 due to the CMS Hospital Readmission Reduction Program implemented in October 2012 [31].
Related to our hypotheses, we found that post-surgical pain at hospital discharge was associated with a higher risk of 30-day readmission following both hip and knee joint arthroplasty. The odds of 30-day readmission was decreased in patients discharged to the community for both hip and knee arthroplasty.
Previous studies examining 30-day readmission for patients with total hip arthroplasty and total knee arthroplasty have consistently reported a cluster of variables associated with readmission. A meta-analysis by Ramkumar and colleagues [32] identified five classifications, including reasons for readmission. The classifications include a) Thromboembolic Disease, b) Joint-Specific, c) Sequelae, d) Cardiac Dysrhythmia and e) Surgical Site Infection. Each classification contains conditions or disorders associated with pain. Along this line, Clement and colleagues reported [16] that pain is one of the common readmitting diagnoses (i.e., ICD-9, 719.45 Pain in joint, pelvic region and thigh; and 729.5 Pain in a limb) among patients with total hip arthroplasty. As noted previously, clinically meaningful measures of pain are not included in Medicare files or other datasets commonly available to investigators.
The American Pain Society and the National Academy of Sciences both recommend that healthcare providers assess and quantify the level of pain, to improve health outcomes [33–35]. In 2001, the Joint Commission created its Pain Management Standard, which requires healthcare providers to ask every patient about pain [35, 36]. This is an important step in identifying pain, but has not been operationalized into a clinical standard that could be integrated in CMS or other files. In our study, pain was recorded using a standardized measure (visual analogue scale) collected in hospitals and included in the Optum database. The emergence of EHRs as a research resource and their aggregation into data warehouses such as Optum provides the opportunity to address questions and issues not possible in Medicare and other administrative data. The EHR also allow access to patient populations not available in Medicare files, e.g., person under 65 years. In addition, advanced data analytics (natural language processing) allow us to extract information on pain and other variables from text (e.g., clinical notes) and other unstructured data. Using the information from pain at hospital discharge, our study was able to examine the risk of 30-day readmission across pain levels in ways not previously feasible.
Limitations
There are challenges in using EHR data in health outcomes research. Data derived from the EHR may be incomplete for some variables, in contrast to administrative data which have fewer missing observations. We excluded patients who did not have a numeric pain value recorded while in the hospital. Of those we excluded (26.6%), 0.6% had a non-numeric or out-of-range pain value from the clinical measurements; 7.0% had a non-numeric value from the signs, diseases and symptoms table of the EHR; and 19.0% had no reported pain value. In addition, we only had a number for pain in the EHR, meaning that the pain score might not have come from a visual analogue scale. Non-numeric pain information was not separated out by surgery type. BMI was treated in a similar manner. In the final step of cohort creation, 0.7% of patients were excluded because they did not have a BMI value recorded between 30 days prior to the surgery and hospital discharge.
In EHR data, the structured ratings of various clinical items can vary substantially across different healthcare systems. In addition, we did not examine variations across healthcare systems and regions. For instance, we did not include the integrated delivery network information or 3-digit zip code as covariates because we did not consider those two variables as valid sources for examining variation. The study EHR data also did not include the educational attainment of patients, functional status or postoperative pain management information; thus, the regression models did not control for these variables.
The rates of readmission might be underestimated if patients were readmitted to a hospital system not in the Optum EHR data network. In addition, while community discharge indicates a discharge to the community from a hospital, it is not clear if patients actually went home. Lastly, the study cohort was selected by ICD-9 and ICD-10 codes and a few CPT codes. Since the Optum EHR database contains various CPT codes which also can identify patients with total knee and hip arthroplasty, future studies need to validate the study cohort using CPT codes.
Conclusions
We found a higher risk of 30-day readmissions following total hip and knee joint arthroplasty surgery in patients reporting high levels of pain (none vs. pain scores 4–10) at hospital discharge. The difference in readmission rates remained after adjusting for patient characteristics. Our study demonstrated the feasibility of using EHR data to study the association of a hard-to-measure variable – pain – with 30-day readmission. Future studies are needed to explore the association of hospital readmissions and other hard-to-measure clinical, social, behavioral and environmental variables captured in EHR data.
Highlights.
Pain results in a higher risk of 30-day readmissions following total arthroplasty
This study highlights the feasibility of using electronic health record data
A hard to measure variable - pain - which is limited in administrative claims data
Acknowledgement
Sarah Toombs Smith, PhD, ELS (University of Texas Medical Branch), provided assistance in proofreading and editing the manuscript. She was not compensated for her contribution. Funding: This study was supported with funding from the National Institutes of Health (R01-AG033134; P2C HD065702; R01-HD069443; K01-HD086290; P30-AG024832, K12 HD055929, 1UL1TR001439).
Appendix Table A.1.
ICD-9, ICD-10, and CPT codes for identifying the study subjects.
| ICD-9 | ICD-10 | CPT |
|---|---|---|
| 81.51 | 0SR90J9, Replacement of Right Hip Joint with Synthetic Substitute, Cemented, Open Approach | 27130, Arthroplasty, acetabular and proximal femoral prosthetic replacement (total hip arthroplasty), with or without autograft or allograft |
| 0SR90JA, Replacement of Right Hip Joint with Synthetic Substitute, Uncemented, Open Approach | ||
| 0SR90JZ, Replacement of Right Hip Joint with Synthetic Substitute, Open Approach | ||
| 0SRB0J9, Replacement of Left Hip Joint with Synthetic Substitute, Cemented, Open Approach | ||
| 0SRBOJA, Replacement of Left Hip Joint with Synthetic Substitute, Uncemented, Open Approach | ||
| 0SRBOJZ, Replacement of Left Hip Joint with Synthetic Substitute, Open Approach | ||
| 81.54 | 0SRC07Z, Replacement of Right Knee Joint with Autologous Tissue Substitute, Open Approach | 27447, Arthroplasty, knee, condyle and plateau; medial AND lateral compartments with or without patella resurfacing (total knee arthroplasty) |
| 0SRCOJZ, Replacement of Right Knee Joint with Synthetic Substitute, Open Approach | ||
| 0SRCOKZ, Replacement of Right Knee Joint with Nonautologous Tissue Substitute, Open Approach | ||
| 0SRCOLZ, Replacement of Right Knee Joint with Medial Unicondylar Synthetic Substitute, Open Approach | ||
| 0SRD07Z, Replacement of Left Knee Joint with Autologous Tissue Substitute, Open Approach | ||
| 0SRDOJZ, Replacement of Left Knee Joint with Synthetic Substitute, Open Approach | ||
| 0SRDOKZ, Replacement of Left Knee Joint with Nonautologous Tissue Substitute, Open Approach | ||
| 0SRDOLZ, Replacement of Left Knee Joint with Medial Unicondylar Synthetic Substitute, Open Approach | ||
| 0SRT07Z, Replacement of Right Knee Joint, Femoral Surface with Autologous Tissue Substitute, Open Approach | ||
| 0SRTOJZ, Replacement of Right Knee Joint, Femoral Surface with Synthetic Substitute, Open Approach | ||
| 0SRTOKZ, Replacement of Right Knee Joint, Femoral Surface with Nonautologous Tissue Substitute, Open Approach | ||
| 0SRU07Z, Replacement of Left Knee Joint, Femoral Surface with Autologous Tissue Substitute, Open Approach | ||
| 0SRUOJZ, Replacement of Left Knee Joint, Femoral Surface with Synthetic Substitute, Open Approach | ||
| 0SRUOKZ, Replacement of Left Knee Joint, Femoral Surface with Nonautologous Tissue Substitute, Open Approach | ||
| 0SRV07Z, Replacement of Right Knee Joint, Tibial Surface with Autologous Tissue Substitute, Open Approach | ||
| 0SRVOJZ, Replacement of Right Knee Joint, Tibial Surface with Synthetic Substitute, Open Approach | ||
| 0SRVOKZ, Replacement of Right Knee Joint, Tibial Surface with Nonautologous Tissue Substitute, Open Approach | ||
| 0SRW07Z, Replacement of Left Knee Joint, Tibial Surface with Autologous Tissue Substitute, Open Approach | ||
| 0SRWOJZ, Replacement of Left Knee Joint, Tibial Surface with Synthetic Substitute, Open Approach | ||
| 0SRWOKZ, Replacement of Left Knee Joint, Tibial Surface with Nonautologous Tissue Substitute, Open Approach |
Appendix Table A.2.
30-day readmission odds ratios for those with a THA
| Effect | Level | Model IIIa |
|---|---|---|
| Adjusted OR (95% CI) | ||
| Pain | 0 | Ref. |
| 1–3 | 0.98 (0.86, 1.13) | |
| 4–6 | 1.23 (1.07, 1.42) | |
| 7–10 | 1.58 (1.31, 1.91) | |
| BMI | Normal | Ref. |
| Underweight | 1.19 (0.81, 1.74) | |
| Overweight | 1.02 (0.91, 1.15) | |
| Obese | 1.22 (1.09, 1.35) | |
| Age | <55 | Ref. |
| 55–64 | 0.94 (0.83, 1.06) | |
| 65–74 | 0.87 (0.77, 0.98) | |
| 75+ | 1.18 (1.04, 1.35) | |
| Gender | Male | Ref. |
| Female | 1.02 (0.94, 1.10) | |
| Race | Caucasian | Ref. |
| African American | 0.90 (0.78, 1.05) | |
| Asian | 1.04 (0.57, 1.92) | |
| Other/Unknown | 0.77 (0.59, 0.99) | |
| Ethnicity | Not Hispanic | Ref. |
| Hispanic | 0.91 (0.67, 1.25) | |
| Unknown | 0.78 (0.62, 0.99) | |
| Charlson Comorbidity | 0 | Ref. |
| 1 | 1.20 (1.08, 1.33) | |
| 2 | 1.39 (1.24, 1.56) | |
| 3+ | 1.88 (1.70, 2.07) | |
| Length of Stay | 1 | Ref. |
| 2 | 1.24 (1.10, 1.40) | |
| 3 | 1.27 (1.11, 1.45) | |
| 4+ | 1.86 (1.60, 2.17) | |
| Surgery Year | 2011 | Ref. |
| 2012 | 0.65 (0.40, 1.06) | |
| 2013 | 0.60 (0.37, 0.98) | |
| 2014 | 0.62 (0.38, 1.01) | |
| 2015 | 0.62 (0.38, 1.01) | |
| 2016 | 0.64 (0.40, 1.04) | |
| 2017 | 0.62 (0.38, 1.00) | |
| 2018 | 0.54 (0.32, 0.93) | |
| Community Discharge | No | Ref. |
| Yes | 0.62 (0.57, 0.69) | |
| DRG | 470 | Ref. |
| 462 | 0.75 (0.45, 1.26) | |
| 469 | 1.50 (1.24, 1.81) | |
| Other | 1.53 (1.19, 1.96) |
Model III included pain, BMI, age, gender, race, ethnicity, Charlson Comorbidity score, length of hospital stay, surgery year, community discharge, and diagnosis related groups (DRGs).
Appendix Table A.3.
30-day readmission odds ratios for those with a TKA
| Effect | Level | Model IIIa |
|---|---|---|
| Adjusted OR (95% CI) | ||
| Pain | 0 | Ref. |
| 1–3 | 0.99 (0.88, 1.12) | |
| 4–6 | 1.15 (1.01, 1.30) | |
| 7–10 | 1.35 (1.16, 1.57) | |
| BMI | Normal | Ref. |
| Underweight | 1.26 (0.74, 2.15) | |
| Overweight | 0.93 (0.84, 1.04) | |
| Obese | 0.94 (0.85, 1.03) | |
| Age | <55 | Ref. |
| 55–64 | 0.94 (0.85, 1.04) | |
| 65–74 | 0.99 (0.90, 1.09) | |
| 75+ | 1.26 (1.13, 1.40) | |
| Gender | Male | Ref. |
| Female | 0.74 (0.70, 0.78) | |
| Race | Caucasian | Ref. |
| African American | 1.15 (1.04, 1.28) | |
| Asian | 0.82 (0.56, 1.20) | |
| Other/Unknown | 0.71 (0.59, 0.85) | |
| Ethnicity | Not Hispanic | Ref. |
| Hispanic | 1.08 (0.90, 1.31) | |
| Unknown | 0.90 (0.76, 1.07) | |
| Charlson Comorbidity | 0 | Ref. |
| 1 | 1.15 (1.06, 1.24) | |
| 2 | 1.32 (1.21, 1.44) | |
| 3+ | 1.80 (1.68, 1.94) | |
| Length of Stay | 1 | Ref. |
| 2 | 1.17 (1.05, 1.30) | |
| 3 | 1.14 (1.02, 1.28) | |
| 4+ | 1.56 (1.38, 1.77) | |
| Surgery Year | 2011 | Ref. |
| 2012 | 1.30 (0.75, 2.23) | |
| 2013 | 1.30 (0.76, 2.23) | |
| 2014 | 1.26 (0.74, 2.16) | |
| 2015 | 1.42 (0.83, 2.43) | |
| 2016 | 1.36 (0.79, 2.33) | |
| 2017 | 1.05 (0.61, 1.80) | |
| 2018 | 1.02 (0.58, 1.81) | |
| Community Discharge | No | Ref. |
| Yes | 0.75 (0.70, 0.81) | |
| DRG | 470 | Ref. |
| 462 | 0.68 (0.58, 0.79) | |
| 469 | 1.50 (1.28, 1.76) | |
| Other | 1.32 (1.08, 1.63) |
Model III included pain, BMI, age, gender, race, ethnicity, Charlson Comorbidity score, length of hospital stay, surgery year, community discharge, and diagnosis related groups (DRGs).
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Burton R Improving care transitions: better coordination of patient transfers among care sites and the community could save money and improve the quality of care. RWJ Health Policy Brief 2012. [Google Scholar]
- 2.Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009;360(14):1418–28. [DOI] [PubMed] [Google Scholar]
- 3.Medicare Payment Advisory Commission. Report to the Congress: promoting greater efficiency in Medicare. 2007. URL: http://www.medpac.gov/docs/default-source/reports/Jun07_EntireReport.pdf.
- 4.Fischer C, Lingsma HF, Marang-van de Mheen PJ, Kringos DS, Klazinga NS, Steyerberg EW. Is the readmission rate a valid quality indicator? a review of the evidence. PloS One 2014;9(11):e112282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Centers for Medicare and Medicaid Services. Comprehensive care for joint replacement model. 2016. URL: https://innovation.cms.gov/initiatives/cjr.
- 6.Centers for Medicare and Medicaid Services. Readmissions reduction program. Department of Health and Human Services: Baltimore, MD: 2015. [Google Scholar]
- 7.Avram V, Petruccelli D, Winemaker M, de Beer J. Total joint arthroplasty readmission rates and reasons for 30-day hospital readmission. J Arthroplasty 2014;29(3):465–8. [DOI] [PubMed] [Google Scholar]
- 8.Keeney JA, Nam D, Johnson SR, Nunley RM, Clohisy JC, Barrack RL. The Impact of risk reduction initiatives on readmission: THA and TKA readmission rates. J Arthroplasty 2015;30(12):2057–60. [DOI] [PubMed] [Google Scholar]
- 9.Cram P, Lu X, Kaboli PJ, Vaughan-Sarrazin MS, Cai X, Wolf BR, et al. Clinical characteristics and outcomes of Medicare patients undergoing total hip arthroplasty, 1991–2008. JAMA 2011;305(15):1560–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Cram P, Lu X, Kates SL, Singh JA, Li Y, Wolf BR. Total knee arthroplasty volume, utilization, and outcomes among Medicare beneficiaries, 1991–2010. JAMA 2012;308(12):1227–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kurtz SM, Lau EC, Ong KL, Adler EM, Kolisek FR, Manley MT. Hospital, patient, and clinical factors Influence 30- and 90-day readmission after primary total hip arthroplasty. J Arthroplasty 2016;31(10):2130–8. [DOI] [PubMed] [Google Scholar]
- 12.Lavernia CJ, Villa JM. Readmission rates in total hip arthroplasty: a granular analysis? J Arthroplasty 2015;30(7):1127–31. [DOI] [PubMed] [Google Scholar]
- 13.Mesko NW, Bachmann KR, Kovacevic D, LoGrasso ME, O’Rourke C, Froimson MI. Thirty-day readmission following total hip and knee arthroplasty - a preliminary single institution predictive model. J Arthroplasty 2014;29(8):1532–8. [DOI] [PubMed] [Google Scholar]
- 14.Oronce CI, Shao H, Shi L. Disparities in 30-day readmissions after total hip arthroplasty. Med Care 2015;53(11):924–30. [DOI] [PubMed] [Google Scholar]
- 15.Paxton EW, Inacio MC, Singh JA, Love R, Bini SA, Namba RS. Are there modifiable risk factors for hospital readmission after total hip arthroplasty in a US healthcare system? Clin Orthop Relat Res 2015;473(11):3446–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Clement RC, Derman PB, Graham DS, Speck RM, Flynn DN, Levin LS, et al. Risk factors, causes, and the economic implications of unplanned readmissions following total hip arthroplasty. J Arthroplasty 2013;28:7–10. [DOI] [PubMed] [Google Scholar]
- 17.Pugely AJ, Callaghan JJ, Martin CT, Cram P, Gao Y. Incidence of and risk factors for 30-day readmission following elective primary total joint arthroplasty: analysis from the ACS-NSQIP. J Arthroplasty 2013;28(9):1499–504. [DOI] [PubMed] [Google Scholar]
- 18.Raines BT, Ponce BA, Reed RD, Richman JS, Hawn MT. Hospital acquired conditions are the strongest predictor for early readmission: an analysis of 26,710 arthroplasties. J Arthroplasty 2015;30(8):1299–307. [DOI] [PubMed] [Google Scholar]
- 19.Saucedo JM, Marecek GS, Wanke TR, Lee J, Stulberg SD, Puri L. Understanding readmission after primary total hip and knee arthroplasty: who’s at risk? J Arthroplasty 2014;29(2):256–60. [DOI] [PubMed] [Google Scholar]
- 20.Schairer WW, Vail TP, Bozic KJ. What are the rates and causes of hospital readmission after total knee arthroplasty? Clin Orthop Relat Res 2014;472(1):181–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Singh JA, Inacio MC, Namba RS, Paxton EW. Rheumatoid arthritis is associated with higher ninety-day hospital readmission rates compared to osteoarthritis after hip or knee arthroplasty: a cohort study. Arthritis Care Res 2015;67(5):718–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kumar A, Karmarkar A, Downer B, Vashist A, Adhikari D, Al Snih S, et al. Current risk adjustment and comorbidity index underperformance in predicting post-acute utilization and hospital readmissions after joint replacements: implications for comprehensive care for joint replacement model. Arthritis Care Res 2017;69(11):1668–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Slocum C, Gerrard P, Black-Schaffer R, Goldstein R, Singhal A, DiVita MA, et al. Functional status predicts acute care readmissions from inpatient rehabilitation in the stroke population. PloS One 2015;10(11):e0142180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kogan E, Twyman K, Heap J, Milentijevic D, Lin JH, Alberts M. Assessing stroke severity using electronic health record data: a machine learning approach. BMC Med Inform Decis Mak 2020;20(1):8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Nunes AP, Yang J, Radican L, Engel SS, Kurtyka K, Tunceli K, et al. Assessing occurrence of hypoglycemia and its severity from electronic health records of patients with type 2 diabetes mellitus. Diabetes Res Clin Pract 2016;121:192–203. [DOI] [PubMed] [Google Scholar]
- 26.RTI International. Measure specifications for measures adopted in the FY 2017 IRF QRP Final Rule. 2016. URL: https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/IRF-Quality-Reporting/Downloads/Measure-Specifications-for-FY17-IRF-QRP-Final-Rule.pdf.
- 27.D’Hoore W, Bouckaert A, Tilquin C. Practical considerations on the use of the Charlson comorbidity index with administrative data bases. J Clin Epidemiol 1996;49(12):1429–33. [DOI] [PubMed] [Google Scholar]
- 28.Ryan SP, Plate JF, Goltz DE, Attarian DE, Wellman SS, Seyler TM, Jiranek WA. Should Medical Severity-Diagnosis Related Group Classification be utilized for reimbursement? An analysis of Elixhauser comorbidities and cost of care. J Arthroplasty 2019;34(7):1312–16. [DOI] [PubMed] [Google Scholar]
- 29.Centers for Disease Control Prevention. Defining adult overweight and obesity. 2017. URL: https://www.cdc.gov/obesity/adult/defining.html.
- 30.Centers for Medicare and Medicaid Services. Hospital-level 30-Day all-cause risk-standardized readmission rate following elective primary total hip arthroplasty (THA) and/or total knee arthroplasty (TKA). 2012. URL: https://cmit.cms.gov/CMIT_public/ViewMeasure?MeasureId=899.
- 31.Ramaswamy A, Marchese M, Cole AP, Harmouch S, Friedlander D, Weissman JS, et al. Comparison of hospital readmission after total hip and total knee arthroplasty vs spinal surgery after implementation of the hospital readmissions reduction program. JAMA Network Open 2019;2(5):e194634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Ramkumar PN, Chu CT, Harris JD, Athiviraham A, Harrington MA, White DL, et al. Causes and rates of unplanned readmissions after elective primary total joint arthroplasty: a systematic review and meta-analysis. Am J Orthop (Belle Mead NJ) 2015;44(9):397–405. [PubMed] [Google Scholar]
- 33.American Pain Society. Principles of analgesic use in the treatment of acute pain and cancer pain. American Pain Society. Clin Pharm 1987;6(7):523–32. [PubMed] [Google Scholar]
- 34.Institute of Medicine Committee on Pain Disability, and Chronic Illness Behavior Pain and disability: clinical, behavioral, and public policy perspectives. Washington (DC): National Academies Press (US) 1987. [PubMed] [Google Scholar]
- 35.Phillips DM. JCAHO pain management standards are unveiled. Joint Commission on Accreditation of Healthcare Organizations. JAMA 2000;284(4):428–9. [DOI] [PubMed] [Google Scholar]
- 36.Committee on Advancing Pain Research Care and Education. Relieving pain in america: a blueprint for transforming prevention, care, education, and research. National Academies Press; 2011. [PubMed] [Google Scholar]