Abstract
Background: Specialty palliative care (SPC) provides patient-centered care to people with serious illness and may reduce costs. Specific cost-saving functions of SPC remain unclear.
Objectives: (1) To assess the effect of SPC on inpatient costs and length of stay (LOS) and (2) to evaluate differences in costs by indication and timing of SPC.
Design: Case–control with in patients who received an SPC consultation and propensity matched controls.
Setting: One large U.S. integrated delivery finance system.
Measurements: Using administrative data, we assessed costs associated with inpatient stays, a subset of whom received an SPC consultation. Consultations were stratified by reasons based on physician discretion: goals of care, pain management, hospice evaluation, nonpain symptom management, or support. The primary outcome was total operating costs and the secondary outcome was hospital LOS.
Results: In total, 1404 patients with SPC consultations associated with unique hospital encounters were matched with 2806 controls. Total operating costs were lower for patients who received an SPC consultation when the consultation was within 0 to 1 days of admission ($6,924 vs. $7,635, p = 0.002). Likewise, LOS was shorter (4.3 vs. 4.7 days, p < 0.001). Upon stratification by reason, goals-of-care consultations early in the hospital stay (days 0–1) were associated with reduced total operating costs ($7,205 vs. $8,677, p < 0.001). Costs were higher for pain management consultations ($7,727 vs. $6,914, p = 0.047). Consultation for hospice evaluation was associated with lower costs, particularly when early (hospital days 0–1: $4,125 vs. $7,415, p < 0.001).
Conclusions: SPC was associated with significant cost saving and decreased LOS when occurring early in a hospitalization and used for goals-of-care and hospice evaluation.
Keywords: goals of care, hospice evaluation, inpatient palliative care, pain management, propensity matching, psychosocial support, symptom management
Introduction
Palliative care has been shown to improve quality of life and symptom burden, while simultaneously reducing costs in certain settings, including for inpatient hospital costs.1–3 This evidence is particularly strong when palliative care is delivered by specialty-level palliative care clinicians.4 In the inpatient setting, cost savings are greater when specialty palliative care (SPC) consultations are earlier in the hospital stay, while potentially allowing more opportunity for goal-concordant care near the end of life, including discharge to hospice.5–7 Furthermore, reduced length of stay (LOS) may be a primary factor in palliative care cost savings.8 However, these studies do not account for the diverse roles of SPC services, including pain and symptom management, goals-of-care discussion, or psychosocial patient and family support; this lack of nuance introduces difficulty when considering development and implementation of palliative care interventions in the context of a growing SPC workforce shortage.9,10
Given that palliative care is a patient-centered intervention that seeks to address the varied array of suffering experienced by individuals with serious illness, patients may receive different elements of palliative care (e.g., symptom management and goals-of-care elicitation). Yet, the traditional method of analyzing the effects of palliative care interventions assumes that individual elements are similar enough to be amalgamated into one uniform “intervention.” This approach is insufficient to understand whether specific elements within palliative care are associated with impacts on outcomes.11,12
One important piece to understand differential effects of SPC is assessing whether and how different functions of SPC consultations are associated with cost savings and utilization. Using administrative data from one academic medical center, we sought to (1) assess the effect of SPC consultation on inpatient total operating costs and LOS and (2) assess differences in costs by type of SPC consultation and time of consultation, where types of consultations were classified as goals of care, pain management, symptom management, hospice evaluation, and psychosocial support.
Methods
Data and sample
We used administrative data from three teaching hospitals (1575 beds) in a U.S. academic medical system to assess the association between SPC consultations and costs during inpatient stays among hospital encounters from July 1, 2016, to December 31, 2017. This work was approved by the UPMC Quality Review Committee.
Reason for consultation
Reasons for SPC consultation are template and designated by the referring clinician at the time the consultation is called. Consulting clinicians select the most pertinent reason for referral from five mutually exclusive categories: goals of care, pain management, hospice evaluation, nonpain symptom management, and support (psychosocial support for patients struggling with being hospitalized or their families). Selection of these categories is intended for clinical utility in communicating the referring clinicians' perceptions of patients' needs to the SPC consultation service. Although an imperfect metric for research, evaluation, and clinical practice, systemic variation in selection of these categories provides a picture of real-time clinical assessment, and how referring clinicians prioritize aspects of patient care when consulting SPC.
Timing
We present the distribution of hospital days on which consultations occurred, measured as whole days, 0 to 10. We excluded consultations that occurred after hospital day 10 (9% of the total) due to difficulties in matching controls. Timing of consultations was also stratified by reason for consultation.
Outcomes
Primary outcome
The primary outcome was total unit operating costs, composed of the following cost categories: supply, drug, direct blood, regular salary, overtime salary, benefits, central benefits, unit operating, and depreciation.2
Secondary outcomes
The secondary outcomes were LOS and variable costs. Variable costs were composed of 100% of direct supplies, direct drugs, direct blood, unit supplies, unit drugs, 60% of regular salaries, overtime salaries, benefits and central benefits, 70% of unit operating, and 0% of depreciation. These criteria are aimed to eliminate costs that cannot be controlled by the facility such as depreciation, utilities, or minimum staffing when the census is 0.
For all cost outcomes, cost estimates are actual costs to the hospital incurred during patient care, not charges billed to the patient.
Exploratory outcomes
As an exploratory analysis, we stratified outcomes by the reason that the primary clinical service called for an SPC consultation.
Statistical analysis
Analyses were conducted using a hospital perspective. Patients were matched using Mahalanobis distance on age, median income, gender, marital status, surgeon versus nonsurgeon attending physician, the Elixhauser comorbidity index, admission priority (nonelective, elective, or trauma), a diagnosis of metastatic cancer, insurance type, severity of illness (calculated using standard recommended techniques for All Patient Refined Diagnosis Related Groups [APR-DRGs]; see Appendix Table A1 for complete list), and inpatient stays in the previous 12 months.13–15 An inclusive match included exact matches for transfer status and ICU stay; a more precise match was run as a sensitivity analysis that assessed for bias in the primary match due to clinical factors and included exact matches for hospital, DRG code, transfer status, and ICU stay.
Controls were also required to have an LOS equal to or greater than the days to consultation for the corresponding case. For example, if the consultation occurred on a patient's fifth day in the hospital, the control must have stayed at least five days as well. Requiring controls to have a LOS at least to the day of consultation for the match mitigates immortal time bias, in which observations must be event free (e.g., survive without discharge) before the time of potential intervention (e.g., consultation).16 Up to three controls were selected per case. Consultations after 10 days were not included. General estimating equation gamma models with log links were constructed for consultations performed within 0 to 1 days of hospital admission, 2 to 3, 4 to 5, and 6 to 10 days.
Unlike similar studies, we included in-hospital decedents in all analyses. We ran a sensitivity analysis with cases and controls matched on in-hospital death to assess whether the matches adequately accounted for severity and acuity of illness. Statistical analyses were conducted using Stata 15.1 (College Station, TX).
Results
Sample
SPC consultations were found for 4711 unique hospital encounters between July 1, 2016, and December 31, 2017, of which 4310 occurred within 10 days of admission. In the inclusive primary match, 4111 consultations were matched to 23,793 controls. In sensitivity analysis with more exact matching, 1404 SPC consultation cases (∼30%) were matched with 2806 controls. Patients who received SPC consultations on average were slightly older than their matched controls (67.5 years vs. 66.1 years, standard mean difference (SMD) = 0.094). Patients receiving consultations were less often Black (16.7% vs. 14.5.0%, SMD = 0.06). Patients who received consultations had a higher median yearly income based on ZIP code from United States Census data ($46,842 vs. $45,111; p < 0.001) (Table 1).
Table 1.
Characteristics of Palliative Care Consultations and Controls after Matching with Standard Mean Difference
| Characteristic (%) | Inclusive match |
Precise match |
||||
|---|---|---|---|---|---|---|
| Consultation n = 4111 | Matched controls n = 23,793 | SMD | Consultation n = 1404 | Matched controls n = 2806 | SMD | |
| Age (mean) | 66.14 | 67.529 | 0.094 | 70.772 | 69.017 | 0.114 |
| Female | 52.0 | 54.3 | 0.046 | 52.3 | 0.535 | −0.025 |
| Race | ||||||
| White | 78.3 | 76.6 | 0.04 | 77.1 | 77.3 | −0.004 |
| Black | 14.5 | 16.7 | 0.06 | 13.5 | 15.1 | −0.046 |
| Unknown | 5.2 | 4.2 | 0.044 | 6.8 | 5.4 | 0.06 |
| Other | 2.0 | 2.5 | 0.03 | 2.6 | 2.2 | 0.022 |
| Married | 46.3 | 46.2 | 0.002 | 45.3 | 43.8 | 0.031 |
| Median income by ZIP code | $46,842 | $45,111 | 0.13 | $46,805 | $45,104 | 0.123 |
| Insurancea | ||||||
| Medicare | 65.7 | 71.5 | 0.127 | 81.5 | 83.6 | −0.057 |
| Commercial | 18.6 | 1.5 | 0.095 | 10.2 | 8.9 | 0.041 |
| Medicaid | 14.3 | 12.8 | 0.043 | 7.6 | 6.8 | 0.031 |
| Self-pay/other | 1.5 | 0.7 | 0.084 | 0.8 | 0.6 | 0.019 |
| Inpatient stays in the past 12 monthsa | ||||||
| 0 or 1 visit | 56.3 | 66.5 | 21.0 | 68.2 | 70.9 | −0.059 |
| 2–3 visits | 23.6 | 22.1 | 3.6 | 16.9 | 14.7 | 0.061 |
| 4 or more visits | 20.0 | 11.4 | 23.9 | 14.9 | 14.4 | 0.014 |
| Admission prioritya | ||||||
| Nonelective | 85.6 | 88.7 | 0.091 | 85.8 | 85.5 | 0.007 |
| Elective | 9.9 | 07.5 | 0.084 | 7.8 | 7.6 | 0.007 |
| Trauma center | 4.5 | 3.8 | 0.034 | 6.5 | 6.9 | −0.016 |
| Admitted via transfera,b | 38.3 | 38.7 | 0.008 | 44.1 | 46.7 | −0.052 |
| Severity of illness* (APR-DRG) | ||||||
| Extreme | 34.1 | 28.2 | 0.128 | 34.3 | 34.8 | −0.012 |
| Major | 48.1 | 50.3 | 0.045 | 47.8 | 47.5 | 0.006 |
| Moderate | 16.1 | 20.2 | 0.106 | 16.4 | 16.1 | 0.007 |
| Minor | 1.7 | 1.3 | 0.032 | 1.6 | 1.6 | −0.001 |
| Elixhauser index | 5.262 | 4.907 | 0.16 | 5.174 | 5.071 | 0.043 |
| Weight lossa | 26.0 | 15.8 | 25.2 | 14.3 | 11.9 | 0.074 |
| Metastatic cancera | ||||||
| Solid tumor without metastasisa | 42.5 | 35.6 | 0.142 | 22.7 | 17.3 | 0.136 |
| Surgery specialty service | 19.5 | 16.5 | 0.08 | 18.1 | 17.2 | 0.023 |
| ICU staya,b | 37.3 | 37.9 | 0.012 | 42.8 | 44.9 | −0.042 |
| Discharge disposition | ||||||
| Hospice | 21.9% | 1.1% | <0.001* | 25.1% | 1.2% | <0.001* |
| SNF | 16.1% | 19.6% | <0.001* | 18.7% | 19.2% | 0.683 |
| Death | 16.5% | 6.0% | <0.001* | 13.9% | 8.7% | <0.001* |
| Discharge disposition by consultation type (n = 4111) | Goals of care (%) | Pain management (%) | Hospice evaluation (%) | Nonpain symptom management (%) | Support | |
|---|---|---|---|---|---|---|
| In-hospital death |
22.10 |
7.40 |
28.29 |
14.00 |
17.28 |
|
| Hospice |
24.78 |
11.24 |
46.92 |
12.00 |
13.58 |
|
| SNF |
22.42 |
10.58 |
15.31 |
11.43 |
12.96 |
|
| Home | 24.51 | 66.42 | 5.82 | 56.86 | 47.53 | |
Exact matches in precise match.
Exact matches in inclusive match.
p < 0.05.
APR-DRGs, All Patient Refined Diagnosis Related Groups; SNF, skilled nursing facility.
Reason for consultation
Among the 3826 patients who received SPC consultation and were included in the primary match, goals-of-care elicitation (41.2%) and pain management (27.2%) were the primary reasons for referral. Consultations were also called for hospice evaluation (20.2%), nonpain symptom management (6.8%), and support (4.6%).
Time to consultation
On average, SPC consultations occurred on hospital day 2.48 (SD 2.10). Consultations for pain management occurred the earliest, on average, hospital day 1.74 (SD 1.58). Consultations for goals of care (mean hospital day 2.69, SD 2.15), nonpain symptom management (mean hospital day 2.64, SD 2.15), and hospice evaluation (mean hospital day 2.87, SD 2.23) occurred less quickly. Consultations occurred latest for support (mean hospital day 2.94, SD 2.54).
Outcomes
Overall, we found that lower total operating costs were associated with SPC consultation compared with matched controls when the consultation was within the first day of admission ($7,604 vs. $9,196, p < 0.001; Table 2). Likewise, shorter LOS was associated with an SPC consultation on hospital days 0 to 1, when compared with matched controls (5.2 vs. 5.4, p = 0.002). Upon stratification by reason for consultation, consultations for goals of care early in the hospital stay (hospital days 0–1) were associated with reduced total operating costs ($7,778 vs. $9,613, p < 0.001). However, higher total operating costs were associated with consultations for pain management ($8,373 vs. $7,724, p = 0.047). SPC consultation for hospice evaluation was associated with lower total operating costs when the consultation occurred at any point in the hospital stay; again, the association was most pronounced when consultations occurred early (hospital days 0–1: $4,167 vs. $7,754, p < 0.001). There was not a statistically significant difference in total operating costs when consultations were for nonpain symptom management or support (Table 3).
Table 2.
Total Operating Costs, Length of Stay, and Variable Costs by Reason for Consultation (U.S. dollars) for Individual Observations in Inclusive Analysis
| Days to consultation | Total N = 4111 |
||
|---|---|---|---|
| Controls | Consult | p | |
| Total operating costs | |||
| 0–1 | 8196 | 7604 | 0 < 0.001* |
| 2–3 | 11,447 | 11,269 | 0.42 |
| 4–5 | 14,658 | 14,058 | 0.127 |
| 6–10 | 20,834 | 20,697 | 0.812 |
| Length of stay | |||
| 0–1 | 5.4 | 5.2 | 0.002* |
| 2–3 | 6.9 | 6.7 | 0.041 |
| 4–5 | 9.3 | 9.1 | 0.336 |
| 6–10 | 12.7 | 12.5 | 0.543 |
| Variable costs | |||
| 0–1 | 3421 | 3052 | 0.016* |
| 2–3 | 4257 | 4017 | 0.42 |
| 4–5 | 4900 | 5191 | 0.59 |
| 6–10 | 7176 | 7674 | 0.484 |
p < 0.05.
Table 3.
Total Operating Costs, Length of Stay, and Variable Costs by Reason for Consultation (U.S. dollars) for Individual Observations in Inclusive Analysis
| Days to consultation | Reason for consultation |
||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Goals of care n = 1485 (36.1%) |
Pain management n = 1513 (36.8%) |
Hospice evaluation n = 601 (14.6%) |
Nonpain symptom management n = 350 (8.5%) |
Support n = 162 (3.9%) |
|||||||||||
| Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | |
| Total operating costs | |||||||||||||||
| 0–1 | 9613 | 7778 | <0.001* | 7724 | 8373 | 0.001* | 7754 | 4167 | <0.001* | 6689 | 6711.5 | 0.951 | 7651 | 8402 | 0.374 |
| 2–3 | 11,882 | 11,170 | 0.032* | 10,994 | 12,920 | <0.001* | 10,700 | 7582 | <0.001* | 11,031 | 12,050 | 0.238 | 12,545 | 12,206 | 0.765 |
| 4–5 | 14,831 | 13,339 | 0.008* | 15,283 | 17,769 | 0.024* | 13,734 | 10,026 | <0.001* | 12,610 | 14,732 | 0.120 | 15,945 | 19,705 | 0.178 |
| 6–10 | 20,393 | 20,298 | 0.901 | 22,899 | 23,384 | 0.756 | 18,537 | 16,172 | 0.022* | 19,905 | 21,196 | 0.534 | 25,003 | 28,993 | 0.280 |
| Length of stay | |||||||||||||||
| 0–1 | 6.1 | 5.1 | <0.001* | 5.1 | 5.6 | <0.001* | 5.1 | 2.6 | <0.001* | 5 | 5.6 | 0.058 | 6.4 | 6.4 | 0.905 |
| 2–3 | 7.4 | 6.9 | 0.006* | 6.5 | 7.3 | 0.001* | 6 | 4.1 | <0.001* | 6.9 | 7.5 | 0.265 | 8.3 | 7.9 | 0.544 |
| 4–5 | 9.5 | 8.9 | 0.073 | 9.2 | 10.8 | 0.006* | 8.2 | 5.9 | <0.001* | 9.8 | 11.9 | 0.037* | 11.7 | 10.8 | 0.576 |
| 6–10 | 12.7 | 12.8 | 0.788 | 13.1 | 13.1 | 0.919 | 11.4 | 9.3 | <0.001* | 12.8 | 13.6 | 0.546 | 15.6 | 16.4 | 0.706 |
| Variable costs | |||||||||||||||
| 0–1 | 3705 | 3091 | 0.059 | 3391 | 3340 | 0.666 | 3095.9 | 1788 | <0.001* | 3089 | 2972 | 0.621 | 3191 | 2892 | 0.897 |
| 2–3 | 4698 | 4167 | 0.318 | 4018 | 4393 | 0.126 | 3851 | 2968 | <0.001* | 3554 | 4233 | 0.115 | 4578 | 4195 | 0.908 |
| 4–5 | 5122 | 5085 | 0.966 | 5063 | 6148 | 0.053 | 4873 | 4060 | 0.017* | 3977 | 5092 | 0.095 | 5127 | 8073 | 0.776 |
| 6–10 | 6847 | 8113 | 0.348 | 7310 | 7779 | 0.546 | 6391 | 6347 | 0.939 | 6938 | 7851 | 0.411 | 16,066 | 8355 | 0.138 |
p < 0.05.
In stratified analyses of LOS, early SPC consultations for goals of care were associated with a reduction in LOS by about 1 day (consultation on hospital days 0–1: 5.1 days vs. 6.1 days, p < 0.001; Table 2). Longer LOS was associated with consultations for pain management (consultation on hospital days 0–1: 5.6 days vs. 5.1 days, p < 0.001). Consultations for hospice evaluation were also associated with shorter LOSs, regardless of when the consultation occurred (at hospital days 0–1: 2.6 vs. 5.1, p < 0.001). There was not a statistically significant difference in LOS when consultations were for nonpain symptom management or psychosocial support (Table 3).
SPC consultations had similarly reduced stratified variable costs, our secondary outcome. Consultations for goals of care on hospital days 0 to 1 were associated with reduced variable costs ($3,052 vs. $3,421, p = 0.016; Table 2). As with total operating costs, SPC consultation for hospice referral was associated with reduced variable costs when the consultation occurred at any point in the hospital stay (hospital days 0–1: $1,788 vs. $3,096, p < 0.001). There was not a statistically significant difference in variable costs when consultations were for nonpain symptom management or support (Table 3).
Sensitivity analysis
In-hospital death
SPC was associated with in-hospital death compared with matched controls (13.9% vs. 8.7%, p < 0.001). The results were generally consistent across the primary match and the sensitivity analysis that also matched on in-hospital death (Appendix Tables A2 and A3).
Precise match
Overall, we found that lower total operating costs were associated with SPC consultation compared with matched controls when the consultation was within the first day of admission ($6,924 vs. $7,635, p = 0.002; Table 4). Likewise, shorter LOS was associated with an SPC consultation on hospital days 0 to 1, when compared with matched controls (4.3 vs. 4.7, p < 0.001). Upon stratification by reason for consultation, consultations for goals of care early in the hospital stay (hospital days 0–1) were associated with reduced total operating costs ($7,205 vs. $8,677, p < 0.001). However, higher total operating costs were associated with consultations for pain management ($7,727 vs. $6,914, p = 0.047). SPC consultation for hospice evaluation was associated with lower total operating costs when the consultation occurred at any point in the hospital stay; again, the association was most pronounced when consultations occurred early (hospital days 0–1: $4,125 vs. $7,415, p < 0.001). There was not a statistically significant difference in total operating costs when consultations were for nonpain symptom management or support (Table 5).
Table 4.
Sensitivity Analysis of Total Operating Costs, Length of Stay, and Variable Costs by Reason for Consultation (U.S. Dollars) for Individual Observations in Precise Analysis
| Days to consultation | Total N = 1404 |
||
|---|---|---|---|
| Controls | Consult | p | |
| Total operating costs | |||
| 0–1 | 7635 | 6924 | 0.002* |
| 2–3 | 10,869 | 10,325 | 0.146 |
| 4–5 | 13,823 | 13,437 | 0.609 |
| 6–10 | 20,073 | 19,290 | 0.515 |
| Length of stay | |||
| 0–1 | 4.7 | 4.3 | <0.001* |
| 2–3 | 6.3 | 6.0 | 0.097 |
| 4–5 | 8.1 | 8.1 | 0.954 |
| 6–10 | 11.9 | 11.6 | 0.613 |
| Variable costs | |||
| 0–1 | 3299 | 3284 | 0.911 |
| 2–3 | 4549 | 4474 | 0.703 |
| 4–5 | 5962 | 6197 | 0.576 |
| 6–10 | 8901 | 9326 | 0.534 |
p < 0.05.
Table 5.
Sensitivity Analysis of Total Operating Costs, Length of Stay, and Variable Costs by Reason for Consultation (U.S. Dollars) for Individual Observations in Precise Analysis
| Days to consultation | Reason for consultation |
||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Goals of care n = 578 (41.2%) |
Pain management n = 382 (27.2%) |
Hospice evaluation n = 284 (20.2%) |
Nonpain symptom management n = 95 (6.8%) |
Support n = 65 (4.6%) |
|||||||||||
| Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | |
| Total operating costs | |||||||||||||||
| 0–1 | 8677 | 7205 | <0.001* | 6914 | 7727 | 0.047* | 7415 | 4125 | <0.001* | 6582 | 6250 | 0.674 | 6988 | 8142 | 0.390 |
| 2–3 | 11,303 | 10,818 | 0.385 | 10,633 | 12,477 | 0.068 | 9901 | 7317 | <0.001* | 10,411 | 9012 | 0.278 | 12,157 | 10,899 | 0.459 |
| 4–5 | 13,704 | 13,383 | 0.754 | 14,433 | 18,493 | 0.177 | 14,538 | 9773 | <0.001* | 8766 | 12,759 | 0.129 | 15,488 | 24,800 | 0.171 |
| 6–10 | 18,413 | 20,109 | 0.323 | 26,372 | 22,710 | 0.409 | 19,606 | 15,392 | 0.012* | 20,951 | 18,423 | 0.553 | 21,393 | 24,476 | 0.610 |
| Length of stay | |||||||||||||||
| 0–1 | 5.2 | 4.3 | <0.001* | 4.4 | 4.9 | 0.024* | 4.7 | 2.3 | <0.001* | 4.6 | 4.6 | 0.943 | 4.2 | 4.8 | 0.416 |
| 2–3 | 6.5 | 6.3 | 0.492 | 6.0 | 7.1 | 0.014* | 5.7 | 3.9 | <0.001* | 6.2 | 6.0 | 0.779 | 7.3 | 6.3 | 0.304 |
| 4–5 | 8.0 | 8.8 | 0.206 | 8.3 | 10.2 | 0.127 | 8.2 | 5.8 | <0.001* | 7.2 | 8.0 | 0.579 | 7.9 | 9.6 | 0.503 |
| 6–10 | 11.5 | 12.4 | 0.305 | 14.1 | 11.6 | 0.160 | 11.1 | 8.9 | 0.028* | 12.3 | 12.9 | 0.809 | 13.5 | 14.7 | 0.731 |
| Variable costs | |||||||||||||||
| 0–1 | 3823 | 3235 | 0.005* | 3005 | 3829 | <0.001* | 3101 | 1949 | <0.001* | 2773 | 3105 | 0.443 | 2757 | 3652 | 0.185 |
| 2–3 | 4877 | 4646 | 0.413 | 4192 | 5248 | 0.021* | 4121 | 3235 | 0.015* | 4312 | 4366 | 0.939 | 5218 | 4853 | 0.688 |
| 4–5 | 5767 | 6187 | 0.437 | 6593 | 8222 | 0.267 | 6395 | 4390 | 0.008* | 4094 | 5628 | 0.272 | 5699 | 13,708 | 0.065 |
| 6–10 | 8416 | 10,433 | 0.045* | 9784 | 9553 | 0.910 | 9235 | 6948 | 0.044* | 9385 | 9536 | 0.952 | 9301 | 10,903 | 0.634 |
p < 0.05.
SPC consultations had similarly reduced stratified variable costs, our secondary outcome. Consultations for goals of care on hospital days 0 to 1 were associated with reduced variable costs ($3,235 vs. $3,823, p = 0.005; Table 4). Higher variable costs were associated with consultations for pain management compared with matched controls ($3,829 vs. $3,005, p < 0.001), an association that dissipated when consultations occurred later in the hospital stay (hospital days 4–10). As with total operating costs, SPC consultation for hospice referral was associated with reduced variable costs when the consultation occurred at any point in the hospital stay (hospital days 0–1: $1,949 vs. $3,101, p < 0.001). There was not a statistically significant difference in variable costs when consultations were for nonpain symptom management or support (Table 5).
In stratified analyses of LOS, early SPC consultations for goals of care were associated with a reduction in LOS by about 1 day (consultation on hospital days 0–1: 5.2 vs. 4.3 days, p < 0.001; Table 4). Longer LOS was associated with consultations for pain management (consultation on hospital days 0–1: 4.9 days vs. 4.4 days, p = 0.024). Consultations for hospice evaluation were also associated with shorter LOSs, regardless of when the consultation occurred (at hospital days 0–1: 2.3 vs. 4.7, p < 0.001). There was not a statistically significant difference in LOS when consultations were for nonpain symptom management or psychosocial support (Table 5).
Discussion
In this retrospective analysis of inpatient SPC consultation, early SPC was associated with cost savings and shorter LOS. This study is consistent with other national reports of SPC's association with reduced costs, fewer ICU stays, and higher likelihood of hospice referral.2 Unlike prior studies, our analyses included decedents and were stratified by the reason for consultation, adding important nuance about patient-centered variability and utility of SPC. When stratified by consultation type, cost savings were associated with goals-of-care and hospice evaluation consultations.
SPC consultations early in the hospital stay were associated with reduced costs, while allowing for the provision of patient-centered care.17 Our stratified analysis brings additional information about the content of SPC consultations most associated with cost reduction: goals-of-care elicitation and hospice evaluation. Upon consultation, SPC may be providing faster and more robust pain management, earlier hospice referral, and clarification of goals of care that align with less-intensive treatment decisions.1 Higher costs for pain management, specifically, may have been driven by a slight increase in LOS for those patients. The reasons for this are unknown and range from SPC being consulted in more complex pain patients, to the service being more cautious in their discharge of patients after changing opiate doses.6,18
Separate from our findings about costs, understanding the elements of palliative care that are most involved in formal inpatient consultations is essential for palliative care research and practice. Current reports call for more complete understanding of the active ingredients of palliative care so that interventions can focus on those higher value elements of care.19 Researchers have begun investigating both active ingredients and dosing of palliative care.17,20 Although clinical trials for behavioral interventions are difficult to conduct with low risk of bias (e.g., contamination and blinding), randomized controlled trials of palliative care with careful process measurement will be essential.1 Research is also needed to better describe how palliative care is currently being delivered clinically, for example, addressing both reasons for consult and potentially additional services. Understanding the elements of palliative care and how they are implemented clinically will allow for more robust evaluations of outcomes (e.g., utilization and quality of life). Furthermore, understanding individual evidence-based elements of palliative care can help guide incorporation of those elements into standard clinical practice outside SPC (i.e., primary palliative care, or elements of palliative care delivered by clinicians from primary care or nonpalliative care specialties). Such work will be essential to meet the growing needs of the patient population considering current SPC workforce limitations.
This study lends valuable insight into the value and outcomes associated with SPC within one health system. Owing to the SPC workforce shortage, even inpatients settings experience gaps in delivering SPC to patients with high palliative care needs.9,21 Identifying the roles, activities, and outcomes of SPC can guide intervention by non-SPC clinicians with SPC coaching or consultation when appropriate.4
Limitations
As with any retrospective observational study, our findings are limited by the granularity or available data and as a result, in their generalizability. The fact that in-hospital death was more prevalent among those who received SPC consultation may impact LOS, although it is not surprising that consultations were associated with in-hospital death and hospice discharge, given frequent consultation very near death.17 Although we matched on APR-DRGs and severity of illness, the lower costs and shorter LOS for people who received palliative care consultations may be due to earlier death. Beyond APR-DRGs, we do not have additional measures of disability, frailty, or cognitive status. Despite our inability to control for all possible confounders, our results are robust in sensitivity analyses; the analyses that included matching on in-hospital death ameliorate some concern about the primary match capturing severity and acuity of illness.22 We do not have access to daily cost data, so we could not conduct pre–post analyses. Clinician-reported reasons for consult were imperfect, although they provide a picture of real-time clinical assessment. Although hospice enrollment results in some postacute cost, costs associated with hospice are significantly lower than those of additional in-hospital days, particularly at the end of life.2,23,24
Although the controls were not perfectly matched on demographics, due to large sample size, small differences—such as the differences seen in the severity of illness rating—were statistically significant. These imbalances are likely clinically insignificant.
Conclusion
This study adds to existing evidence that inpatient SPC is cost saving, while adding valuable stratification data about the roles of SPC and their respective degrees of cost savings, with maximum cost savings for goals-of-care assessments and hospice evaluation. Furthermore, early SPC is associated with both cost savings and shorter LOSs. Understanding the roles and outcomes of SPC delivery can provide groundwork for effectively meeting the needs of people with serious illness.
Appendix Table A1.
Diagnosis Related Group Matching
| DRG code | Control (%) | Palliative consult (%) | SMD |
|---|---|---|---|
| 1 | 0.48 | 0.92 | 0.053 |
| 2 | 0.04 | 0.07 | 0.015 |
| 3 | 0.46 | 0.90 | 0.053 |
| 4 | 0.14 | 0.29 | 0.033 |
| 6 | 0.01 | 0.02 | 0.009 |
| 11 | 0.36 | 0.39 | 0.005 |
| 12 | 0.48 | 0.54 | 0.008 |
| 13 | 0.12 | 0.12 | 0.000 |
| 14 | 0.07 | 0.15 | 0.024 |
| 16 | 0.10 | 0.17 | 0.020 |
| 17 | 0.03 | 0.05 | 0.012 |
| 20 | 0.12 | 0.15 | 0.008 |
| 23 | 2.00 | 1.58 | 0.032 |
| 24 | 0.05 | 0.07 | 0.011 |
| 25 | 0.32 | 0.49 | 0.026 |
| 26 | 0.08 | 0.15 | 0.018 |
| 27 | 0.01 | 0.02 | 0.009 |
| 28 | 0.03 | 0.07 | 0.017 |
| 29 | 0.06 | 0.10 | 0.015 |
| 31 | 0.01 | 0.02 | 0.009 |
| 32 | 0.01 | 0.02 | 0.009 |
| 40 | 0.13 | 0.24 | 0.025 |
| 41 | 0.03 | 0.05 | 0.012 |
| 52 | 0.04 | 0.05 | 0.003 |
| 54 | 4.36 | 2.63 | 0.095 |
| 55 | 0.84 | 0.66 | 0.021 |
| 56 | 0.15 | 0.22 | 0.017 |
| 57 | 0.20 | 0.29 | 0.019 |
| 62 | 0.01 | 0.02 | 0.009 |
| 64 | 6.61 | 3.65 | 0.135 |
| 65 | 0.32 | 0.54 | 0.033 |
| 66 | 0.03 | 0.05 | 0.012 |
| 68 | 0.01 | 0.02 | 0.009 |
| 70 | 0.12 | 0.22 | 0.025 |
| 71 | 0.01 | 0.02 | 0.009 |
| 73 | 0.03 | 0.05 | 0.012 |
| 74 | 0.06 | 0.12 | 0.019 |
| 82 | 0.31 | 0.24 | 0.013 |
| 83 | 0.04 | 0.07 | 0.015 |
| 84 | 0.03 | 0.05 | 0.012 |
| 85 | 0.99 | 0.80 | 0.020 |
| 86 | 0.39 | 0.56 | 0.025 |
| 87 | 0.07 | 0.10 | 0.010 |
| 91 | 0.20 | 0.27 | 0.015 |
| 92 | 0.07 | 0.12 | 0.016 |
| 94 | 0.02 | 0.07 | 0.027 |
| 97 | 0.05 | 0.05 | 0.001 |
| 98 | 0.01 | 0.02 | 0.009 |
| 100 | 0.36 | 0.46 | 0.016 |
| 101 | 0.03 | 0.05 | 0.007 |
| 103 | 0.01 | 0.02 | 0.009 |
| 115 | 0.00 | 0.02 | 0.017 |
| 121 | 0.01 | 0.02 | 0.009 |
| 125 | 0.01 | 0.02 | 0.009 |
| 129 | 0.03 | 0.05 | 0.012 |
| 130 | 0.01 | 0.02 | 0.009 |
| 133 | 0.01 | 0.02 | 0.009 |
| 134 | 0.01 | 0.02 | 0.009 |
| 146 | 0.16 | 0.17 | 0.004 |
| 147 | 0.71 | 0.41 | 0.040 |
| 148 | 0.01 | 0.02 | 0.009 |
| 150 | 0.00 | 0.02 | 0.017 |
| 154 | 0.01 | 0.02 | 0.009 |
| 157 | 0.19 | 0.15 | 0.012 |
| 158 | 0.11 | 0.12 | 0.005 |
| 163 | 0.50 | 0.46 | 0.005 |
| 164 | 0.03 | 0.05 | 0.012 |
| 166 | 1.01 | 0.95 | 0.007 |
| 167 | 0.10 | 0.20 | 0.026 |
| 168 | 0.01 | 0.02 | 0.009 |
| 175 | 0.14 | 0.20 | 0.013 |
| 176 | 0.21 | 0.32 | 0.020 |
| 177 | 1.74 | 1.51 | 0.018 |
| 178 | 0.71 | 0.80 | 0.010 |
| 179 | 0.01 | 0.02 | 0.009 |
| 180 | 1.32 | 0.97 | 0.032 |
| 181 | 1.11 | 1.07 | 0.004 |
| 182 | 0.04 | 0.07 | 0.015 |
| 183 | 0.01 | 0.02 | 0.009 |
| 184 | 0.05 | 0.07 | 0.011 |
| 186 | 0.23 | 0.32 | 0.016 |
| 187 | 0.04 | 0.07 | 0.015 |
| 189 | 0.57 | 0.56 | 0.001 |
| 190 | 0.16 | 0.29 | 0.028 |
| 191 | 0.17 | 0.27 | 0.020 |
| 192 | 0.01 | 0.02 | 0.009 |
| 193 | 0.85 | 0.90 | 0.005 |
| 194 | 0.38 | 0.56 | 0.026 |
| 196 | 0.37 | 0.39 | 0.002 |
| 199 | 0.03 | 0.07 | 0.017 |
| 200 | 0.04 | 0.10 | 0.021 |
| 204 | 0.03 | 0.05 | 0.012 |
| 205 | 0.18 | 0.32 | 0.027 |
| 206 | 0.08 | 0.15 | 0.021 |
| 207 | 0.37 | 0.37 | 0.000 |
| 208 | 0.61 | 0.63 | 0.002 |
| 215 | 0.01 | 0.02 | 0.009 |
| 216 | 0.03 | 0.07 | 0.019 |
| 219 | 0.04 | 0.07 | 0.015 |
| 226 | 0.03 | 0.05 | 0.012 |
| 228 | 0.03 | 0.05 | 0.012 |
| 233 | 0.01 | 0.02 | 0.009 |
| 235 | 0.03 | 0.05 | 0.012 |
| 239 | 0.08 | 0.17 | 0.026 |
| 240 | 0.06 | 0.10 | 0.014 |
| 242 | 0.04 | 0.07 | 0.015 |
| 243 | 0.01 | 0.02 | 0.009 |
| 246 | 0.03 | 0.05 | 0.012 |
| 247 | 0.04 | 0.07 | 0.015 |
| 248 | 0.03 | 0.05 | 0.012 |
| 250 | 0.01 | 0.02 | 0.012 |
| 252 | 0.27 | 0.39 | 0.021 |
| 253 | 0.11 | 0.17 | 0.015 |
| 254 | 0.01 | 0.02 | 0.009 |
| 260 | 0.04 | 0.07 | 0.013 |
| 264 | 0.29 | 0.34 | 0.010 |
| 266 | 0.04 | 0.07 | 0.015 |
| 270 | 0.26 | 0.34 | 0.015 |
| 271 | 0.04 | 0.07 | 0.015 |
| 273 | 0.01 | 0.02 | 0.009 |
| 280 | 0.48 | 0.61 | 0.018 |
| 281 | 0.11 | 0.17 | 0.018 |
| 283 | 0.08 | 0.15 | 0.018 |
| 286 | 0.79 | 0.75 | 0.005 |
| 287 | 0.36 | 0.44 | 0.012 |
| 291 | 1.92 | 2.21 | 0.021 |
| 292 | 0.43 | 0.63 | 0.028 |
| 299 | 0.41 | 0.46 | 0.008 |
| 300 | 0.49 | 0.58 | 0.013 |
| 301 | 0.07 | 0.10 | 0.010 |
| 303 | 0.01 | 0.02 | 0.009 |
| 304 | 0.03 | 0.05 | 0.012 |
| 305 | 0.03 | 0.05 | 0.012 |
| 306 | 0.03 | 0.05 | 0.007 |
| 308 | 0.40 | 0.46 | 0.010 |
| 309 | 0.06 | 0.12 | 0.023 |
| 312 | 0.01 | 0.02 | 0.009 |
| 313 | 0.04 | 0.07 | 0.015 |
| 314 | 0.79 | 1.00 | 0.022 |
| 315 | 0.09 | 0.20 | 0.027 |
| 326 | 0.45 | 0.51 | 0.009 |
| 327 | 0.11 | 0.24 | 0.032 |
| 328 | 0.01 | 0.02 | 0.009 |
| 329 | 0.76 | 0.92 | 0.018 |
| 330 | 0.45 | 0.68 | 0.030 |
| 331 | 0.05 | 0.07 | 0.011 |
| 335 | 0.03 | 0.05 | 0.012 |
| 336 | 0.03 | 0.07 | 0.017 |
| 337 | 0.01 | 0.02 | 0.009 |
| 344 | 0.01 | 0.02 | 0.009 |
| 345 | 0.04 | 0.05 | 0.003 |
| 348 | 0.01 | 0.02 | 0.009 |
| 353 | 0.01 | 0.02 | 0.009 |
| 356 | 0.16 | 0.27 | 0.022 |
| 357 | 0.09 | 0.17 | 0.023 |
| 358 | 0.01 | 0.02 | 0.009 |
| 368 | 0.03 | 0.07 | 0.019 |
| 371 | 0.37 | 0.37 | 0.002 |
| 372 | 0.08 | 0.15 | 0.018 |
| 374 | 2.11 | 1.31 | 0.061 |
| 375 | 2.66 | 1.65 | 0.069 |
| 377 | 0.38 | 0.54 | 0.023 |
| 378 | 0.10 | 0.20 | 0.024 |
| 380 | 0.04 | 0.10 | 0.021 |
| 381 | 0.11 | 0.12 | 0.005 |
| 385 | 0.01 | 0.02 | 0.009 |
| 386 | 0.01 | 0.02 | 0.009 |
| 388 | 0.66 | 0.56 | 0.012 |
| 389 | 0.43 | 0.54 | 0.015 |
| 390 | 0.01 | 0.02 | 0.009 |
| 391 | 0.76 | 0.85 | 0.011 |
| 392 | 0.45 | 0.68 | 0.031 |
| 393 | 0.37 | 0.54 | 0.025 |
| 394 | 0.36 | 0.58 | 0.033 |
| 395 | 0.01 | 0.02 | 0.009 |
| 405 | 0.03 | 0.05 | 0.012 |
| 406 | 0.08 | 0.15 | 0.021 |
| 408 | 0.10 | 0.12 | 0.008 |
| 409 | 0.01 | 0.02 | 0.009 |
| 415 | 0.03 | 0.05 | 0.012 |
| 418 | 0.01 | 0.02 | 0.012 |
| 420 | 0.11 | 0.17 | 0.015 |
| 421 | 0.10 | 0.17 | 0.020 |
| 423 | 0.03 | 0.05 | 0.012 |
| 424 | 0.05 | 0.05 | 0.001 |
| 432 | 0.59 | 0.63 | 0.005 |
| 433 | 0.06 | 0.12 | 0.021 |
| 435 | 2.45 | 1.80 | 0.045 |
| 436 | 1.53 | 1.12 | 0.036 |
| 437 | 0.02 | 0.05 | 0.015 |
| 438 | 0.03 | 0.05 | 0.012 |
| 439 | 0.01 | 0.02 | 0.009 |
| 441 | 1.61 | 1.36 | 0.020 |
| 442 | 0.40 | 0.58 | 0.027 |
| 443 | 0.01 | 0.02 | 0.009 |
| 444 | 0.11 | 0.17 | 0.016 |
| 445 | 0.18 | 0.24 | 0.015 |
| 446 | 0.01 | 0.02 | 0.009 |
| 453 | 0.01 | 0.02 | 0.012 |
| 456 | 0.02 | 0.07 | 0.024 |
| 457 | 0.06 | 0.07 | 0.007 |
| 460 | 0.01 | 0.02 | 0.009 |
| 463 | 0.03 | 0.05 | 0.012 |
| 464 | 0.11 | 0.17 | 0.015 |
| 467 | 0.03 | 0.05 | 0.012 |
| 469 | 0.01 | 0.02 | 0.009 |
| 470 | 0.01 | 0.02 | 0.009 |
| 475 | 0.03 | 0.05 | 0.012 |
| 477 | 0.03 | 0.05 | 0.012 |
| 478 | 0.18 | 0.22 | 0.009 |
| 479 | 0.01 | 0.02 | 0.009 |
| 480 | 0.12 | 0.20 | 0.019 |
| 481 | 0.19 | 0.29 | 0.020 |
| 482 | 0.01 | 0.02 | 0.009 |
| 483 | 0.00 | 0.02 | 0.017 |
| 492 | 0.04 | 0.10 | 0.021 |
| 493 | 0.11 | 0.15 | 0.009 |
| 500 | 0.01 | 0.02 | 0.012 |
| 511 | 0.01 | 0.02 | 0.009 |
| 515 | 0.09 | 0.12 | 0.010 |
| 516 | 0.03 | 0.05 | 0.012 |
| 518 | 0.02 | 0.05 | 0.018 |
| 535 | 0.01 | 0.05 | 0.021 |
| 536 | 0.07 | 0.12 | 0.016 |
| 537 | 0.00 | 0.02 | 0.017 |
| 539 | 0.07 | 0.10 | 0.010 |
| 540 | 0.01 | 0.02 | 0.012 |
| 542 | 1.01 | 0.66 | 0.039 |
| 543 | 2.78 | 1.34 | 0.102 |
| 544 | 0.01 | 0.02 | 0.009 |
| 545 | 0.05 | 0.12 | 0.024 |
| 549 | 0.00 | 0.02 | 0.017 |
| 551 | 0.22 | 0.29 | 0.015 |
| 552 | 0.17 | 0.27 | 0.020 |
| 553 | 0.01 | 0.02 | 0.009 |
| 554 | 0.03 | 0.05 | 0.012 |
| 555 | 0.02 | 0.05 | 0.018 |
| 556 | 0.03 | 0.05 | 0.012 |
| 558 | 0.03 | 0.05 | 0.012 |
| 559 | 0.02 | 0.05 | 0.018 |
| 560 | 0.03 | 0.05 | 0.012 |
| 563 | 0.05 | 0.10 | 0.017 |
| 564 | 0.01 | 0.02 | 0.012 |
| 565 | 0.03 | 0.05 | 0.007 |
| 570 | 0.03 | 0.05 | 0.007 |
| 579 | 0.02 | 0.05 | 0.015 |
| 580 | 0.08 | 0.12 | 0.013 |
| 582 | 0.01 | 0.02 | 0.009 |
| 592 | 0.01 | 0.02 | 0.009 |
| 597 | 0.02 | 0.05 | 0.018 |
| 598 | 0.10 | 0.17 | 0.019 |
| 600 | 0.03 | 0.05 | 0.012 |
| 602 | 0.12 | 0.15 | 0.008 |
| 603 | 0.06 | 0.12 | 0.019 |
| 605 | 0.01 | 0.02 | 0.009 |
| 606 | 0.00 | 0.02 | 0.017 |
| 607 | 0.06 | 0.07 | 0.004 |
| 615 | 0.01 | 0.02 | 0.009 |
| 622 | 0.06 | 0.07 | 0.004 |
| 623 | 0.01 | 0.02 | 0.009 |
| 637 | 0.08 | 0.12 | 0.012 |
| 638 | 0.01 | 0.02 | 0.009 |
| 640 | 0.75 | 0.85 | 0.011 |
| 641 | 0.59 | 0.71 | 0.015 |
| 642 | 0.03 | 0.07 | 0.017 |
| 643 | 0.19 | 0.22 | 0.007 |
| 644 | 0.06 | 0.12 | 0.021 |
| 657 | 0.04 | 0.07 | 0.015 |
| 668 | 0.01 | 0.02 | 0.012 |
| 669 | 0.01 | 0.02 | 0.009 |
| 673 | 0.09 | 0.17 | 0.023 |
| 674 | 0.03 | 0.07 | 0.019 |
| 682 | 1.09 | 1.24 | 0.014 |
| 683 | 0.59 | 0.85 | 0.031 |
| 686 | 0.05 | 0.05 | 0.001 |
| 687 | 0.02 | 0.07 | 0.024 |
| 689 | 0.14 | 0.22 | 0.018 |
| 690 | 0.18 | 0.32 | 0.027 |
| 694 | 0.04 | 0.07 | 0.015 |
| 695 | 0.00 | 0.02 | 0.017 |
| 698 | 0.39 | 0.54 | 0.022 |
| 699 | 0.16 | 0.24 | 0.019 |
| 700 | 0.01 | 0.02 | 0.009 |
| 715 | 0.01 | 0.02 | 0.009 |
| 723 | 0.13 | 0.12 | 0.002 |
| 727 | 0.00 | 0.02 | 0.017 |
| 728 | 0.01 | 0.02 | 0.009 |
| 734 | 0.06 | 0.10 | 0.012 |
| 736 | 0.01 | 0.02 | 0.009 |
| 737 | 0.06 | 0.10 | 0.014 |
| 739 | 0.01 | 0.02 | 0.009 |
| 744 | 0.11 | 0.17 | 0.018 |
| 746 | 0.03 | 0.05 | 0.012 |
| 749 | 0.02 | 0.05 | 0.015 |
| 754 | 0.30 | 0.17 | 0.026 |
| 755 | 0.58 | 0.39 | 0.027 |
| 756 | 0.01 | 0.02 | 0.009 |
| 758 | 0.03 | 0.05 | 0.012 |
| 759 | 0.02 | 0.05 | 0.018 |
| 760 | 0.02 | 0.05 | 0.015 |
| 802 | 0.01 | 0.02 | 0.012 |
| 808 | 0.29 | 0.34 | 0.009 |
| 809 | 0.22 | 0.39 | 0.031 |
| 810 | 0.01 | 0.02 | 0.009 |
| 811 | 0.05 | 0.10 | 0.017 |
| 812 | 0.28 | 0.41 | 0.022 |
| 813 | 0.13 | 0.27 | 0.032 |
| 814 | 0.02 | 0.05 | 0.018 |
| 815 | 0.05 | 0.10 | 0.017 |
| 820 | 0.03 | 0.05 | 0.012 |
| 821 | 0.03 | 0.05 | 0.007 |
| 822 | 0.01 | 0.02 | 0.009 |
| 823 | 0.25 | 0.22 | 0.007 |
| 824 | 0.22 | 0.27 | 0.009 |
| 827 | 0.03 | 0.05 | 0.012 |
| 829 | 0.11 | 0.20 | 0.023 |
| 830 | 0.00 | 0.02 | 0.017 |
| 834 | 1.11 | 0.95 | 0.016 |
| 835 | 0.27 | 0.34 | 0.013 |
| 836 | 0.13 | 0.12 | 0.002 |
| 837 | 0.10 | 0.15 | 0.013 |
| 838 | 0.11 | 0.17 | 0.018 |
| 839 | 0.18 | 0.24 | 0.015 |
| 840 | 0.81 | 0.61 | 0.024 |
| 841 | 0.88 | 0.85 | 0.003 |
| 842 | 0.07 | 0.10 | 0.010 |
| 843 | 0.11 | 0.17 | 0.016 |
| 844 | 0.30 | 0.34 | 0.007 |
| 846 | 0.13 | 0.20 | 0.017 |
| 847 | 0.64 | 0.80 | 0.019 |
| 848 | 0.09 | 0.10 | 0.002 |
| 849 | 0.12 | 0.15 | 0.007 |
| 853 | 1.27 | 1.70 | 0.036 |
| 854 | 0.13 | 0.20 | 0.017 |
| 856 | 0.12 | 0.17 | 0.013 |
| 857 | 0.11 | 0.17 | 0.018 |
| 862 | 0.14 | 0.20 | 0.014 |
| 863 | 0.06 | 0.12 | 0.019 |
| 864 | 0.06 | 0.10 | 0.014 |
| 867 | 0.01 | 0.02 | 0.009 |
| 870 | 0.46 | 0.61 | 0.021 |
| 871 | 7.54 | 6.28 | 0.050 |
| 872 | 0.19 | 0.39 | 0.037 |
| 880 | 0.01 | 0.02 | 0.009 |
| 884 | 0.12 | 0.17 | 0.013 |
| 897 | 0.04 | 0.07 | 0.015 |
| 901 | 0.00 | 0.02 | 0.017 |
| 907 | 0.19 | 0.24 | 0.012 |
| 908 | 0.08 | 0.15 | 0.021 |
| 917 | 0.20 | 0.32 | 0.023 |
| 918 | 0.06 | 0.12 | 0.019 |
| 919 | 0.49 | 0.56 | 0.009 |
| 920 | 0.19 | 0.29 | 0.022 |
| 923 | 0.01 | 0.02 | 0.009 |
| 939 | 0.06 | 0.07 | 0.007 |
| 940 | 0.06 | 0.07 | 0.004 |
| 947 | 1.40 | 0.88 | 0.049 |
| 948 | 5.95 | 2.34 | 0.182 |
| 949 | 0.01 | 0.02 | 0.012 |
| 955 | 0.01 | 0.02 | 0.009 |
| 956 | 0.03 | 0.05 | 0.012 |
| 957 | 0.01 | 0.02 | 0.009 |
| 958 | 0.01 | 0.02 | 0.009 |
| 963 | 0.03 | 0.07 | 0.022 |
| 964 | 0.05 | 0.07 | 0.011 |
| 974 | 0.01 | 0.02 | 0.009 |
| 981 | 1.36 | 1.56 | 0.016 |
| 982 | 0.25 | 0.39 | 0.024 |
| 987 | 0.25 | 0.34 | 0.016 |
| 988 | 0.06 | 0.12 | 0.023 |
p < 0.05.
DRGs, Diagnosis Related Groups.
Appendix Table A2.
Sensitivity Analysis of Total Operating Costs, Length of Stay, and Variable Costs (U.S. Dollars) for Individual Observations in Inclusive Analysis, Matched on Death
| Days to consultation | Full matched sample |
Among in-hospital decedents |
Among survivors to hospital discharge |
||||||
|---|---|---|---|---|---|---|---|---|---|
|
N = 3855 |
N = 496 |
N = 3417 |
|||||||
| Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | |
| Total operating costs | |||||||||
| 0–1 | 7750 | 7590 | 0.19 | 10639 | 8796.1 | 0.008* | 7581 | 7460 | 0.326 |
| 2–3 | 10,961 | 11,315 | 0.114 | 14,847 | 14328.7 | 0.683 | 10,615 | 10,875 | 0.243 |
| 4–5 | 14,240 | 14,145 | 0.819 | 21,494 | 16687.6 | 0.021* | 13,525 | 13,642 | 0.779 |
| 6–10 | 20,495 | 21,195 | 0.265 | 28,012 | 24564.8 | 0.185 | 19,480 | 20,412 | 0.148 |
| Length of stay | |||||||||
| 0–1 | 5.2 | 5.2 | 0.644 | 4.8 | 4.7 | 0.902 | 5.3 | 5.3 | 0.685 |
| 2–3 | 6.7 | 6.7 | 0.968 | 7.4 | 6.7 | 0.089 | 6.6 | 6.7 | 0.507 |
| 4–5 | 9.1 | 9.0 | 0.811 | 9.4 | 8.2 | 0.053 | 9.0 | 9.2 | 0.555 |
| 6–10 | 12.7 | 12.7 | 0.95 | 12.6 | 11.3 | 0.145 | 12.7 | 13 | 0.428 |
| Variable costs | |||||||||
| 0–1 | 3308 | 3025 | 0.092 | 4578 | 4021 | 0.064 | 3235 | 2917 | 0.082 |
| 2–3 | 4172 | 4017 | 0.641 | 5735 | 5450 | 0.51 | 4040 | 3827 | 0.555 |
| 4–5 | 4719 | 5140 | 0.489 | 7637 | 6682 | 0.154 | 4376 | 4875 | 0.457 |
| 6–10 | 7110 | 7684 | 0.48 | 9805 | 9671 | 0.885 | 6779 | 7247 | 0.609 |
Appendix Table A3.
Sensitivity Analysis of Total Operating Costs, Length of Stay, and Variable Costs by Reason for Consultation (U.S. Dollars) for Individual Observations in Inclusive Analysis, Matched on Death
| Days to consultation | Reason for consultation |
||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Full matched sample | |||||||||||||||
| Goals of care |
Pain management |
Hospice evaluation |
Nonpain symptom management |
Support |
|||||||||||
|
n = 1370 (35.5%) |
n = 1463 (38.0%) |
n = 557 (14.5%) |
n = 318 (8.3%) |
n = 147 (3.8%) |
|||||||||||
| Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | |
| Variable costs | |||||||||||||||
| 0–1 | 3504 | 3102 | 0.285 | 3321 | 3294 | 0.82 | 3016 | 1678.9 | <0.001* | 2987 | 3022 | 0.889 | 3218 | 2847 | 0.878 |
| 2–3 | 4530 | 4152 | 0.533 | 3998 | 4411 | 0.099 | 3947 | 2955.2 | <0.001* | 3492 | 4218 | 0.112 | 4430 | 4197 | 0.946 |
| 4–5 | 4749 | 5005 | 0.799 | 5055 | 6049 | 0.082 | 4819 | 4047.5 | 0.031* | 4044 | 5169 | 0.137 | 5207 | 8073 | 0.791 |
| 6–10 | 6727 | 8085 | 0.403 | 7187 | 7839 | 0.437 | 6214 | 6387.2 | 0.777 | 7043 | 8082 | 0.398 | 16,512 | 7958 | 0.117 |
| Length of stay | |||||||||||||||
| 0–1 | 6.0 | 5.2 | <0.001* | 5 | 5.6 | <0.001* | 4.6 | 2.6 | <0.001* | 4.7 | 5.7 | 0.001* | 6.3 | 6.3 | 1 |
| 2–3 | 7.0 | 6.8 | 0.531 | 6.4 | 7.3 | <0.001* | 6 | 4.1 | <0.001* | 7 | 7.6 | 0.306 | 7.8 | 7.9 | 0.961 |
| 4–5 | 9.4 | 8.9 | 0.17 | 8.8 | 10.3 | 0.009* | 7.9 | 5.9 | <0.001* | 9.8 | 12.3 | 0.034* | 10.9 | 10.8 | 0.968 |
| 6–10 | 12.7 | 13 | 0.661 | 12.9 | 13.4 | 0.519 | 11.4 | 9.3 | <0.001* | 13 | 14 | 0.514 | 15.1 | 16.7 | 0.441 |
| Total unit operating costs | |||||||||||||||
| 0–1 | 9021 | 7870 | <0.001* | 7431 | 8279 | <0.001* | 7327 | 4075 | <0.001* | 6314 | 6833 | 0.172 | 7459 | 8356 | 0.308 |
| 2–3 | 11,013 | 11,096 | 0.807 | 10,889 | 12,972 | <0.001* | 10,450 | 7698 | <0.001* | 11,079 | 12,195 | 0.218 | 11,769 | 12,206 | 0.692 |
| 4–5 | 14,126 | 13,274 | 0.145 | 14,761 | 17,689 | 0.008* | 13,688 | 10,129 | <0.001* | 12,920 | 15,309 | 0.119 | 15,754 | 19,705 | 0.149 |
| 6–10 | 19,989 | 20,705 | 0.392 | 22,481 | 24,059 | 0.348 | 18,118 | 16,592 | 0.182 | 20,087 | 22,040 | 0.398 | 23,239 | 28,783 | 0.13 |
| Days to consultation | Among in-hospital decedents |
||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Goals of care |
Pain management |
Hospice evaluation |
Nonpain symptom management |
Support |
|||||||||||
|
n = 246 (49.6%) |
n = 72 (14.5%) |
n = 127 (25.6%) |
n = 27 (5.4%) |
n = 24 (4.8%) |
|||||||||||
| Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | |
| Variable costs | |||||||||||||||
| 0–1 |
4903 |
4199 |
0.097 |
4893 |
5057 |
0.823 |
3582 |
2371 |
0.006* |
4575 |
6263 |
0.457 |
6483 |
4070 |
0.222 |
| 2–3 |
5787 |
5751 |
0.949 |
7694 |
6789 |
0.623 |
5081 |
4025 |
0.116 |
4192 |
6601 |
0.276 |
6318 |
5762 |
0.754 |
| 4–5 |
7678 |
7178 |
0.617 |
13,216 |
8748 |
0.188 |
6013 |
4631 |
0.093 |
5015 |
7273 |
0.412 |
8666 |
10,148 |
0.697 |
| 6–10 |
9709 |
9873 |
0.877 |
12,119 |
11,051 |
0.78 |
7731 |
8248 |
0.772 |
14,278 |
8951 |
0.126 |
8201 |
8945 |
0.898 |
| Length of stay | |||||||||||||||
| 0–1 |
5.2 |
4.4 |
0.04* |
6 |
7.1 |
0.209 |
2.9 |
2.3 |
0.03* |
3.3 |
8 |
0.027* |
6.6 |
7.6 |
0.722 |
| 2–3 |
6.9 |
7 |
0.867 |
8.6 |
9.6 |
0.57 |
6 |
3.6 |
<0.001* |
10.6 |
9.3 |
0.566 |
11.7 |
9 |
0.227 |
| 4–5 |
9.8 |
8 |
0.052 |
12.4 |
13.5 |
0.673 |
7.9 |
5.4 |
0.001* |
7 |
11 |
0.213 |
10.9 |
10.7 |
0.953 |
| 6–10 |
12.6 |
11.6 |
0.348 |
13.7 |
13.5 |
0.936 |
11.8 |
9.3 |
0.1 |
12.9 |
11.6 |
0.74 |
14 |
11.3 |
0.662 |
| Total unit operating costs | |||||||||||||||
| 0–1 |
10,265 |
8975 |
0.133 |
14,989 |
11,944 |
0.146 |
6073 |
5007 |
0.076 |
7085 |
13,227 |
0.115 |
15,132 |
7935 |
0.204 |
| 2–3 |
14,005 |
14,619 |
0.64 |
21,491 |
20,937 |
0.923 |
12,747 |
9817 |
0.119 |
10,392 |
12,537 |
0.716 |
23,093 |
19,933 |
0.799 |
| 4–5 |
21,107 |
16,184 |
0.041* |
29,379 |
32,441 |
0.781 |
21,464 |
11,394 |
0.002* |
14,051 |
14,598 |
0.922 |
17,786 |
20,370 |
0.798 |
| 6–10 | 28,686 | 24,277 | 0.122 | 33,082 | 30,361 | 0.789 | 21,663 | 20,309 | 0.758 | 30,345 | 23,467 | 0.306 | 16,278 | 41,138 | 0.327 |
| Days to consultation | Among survivors to hospital discharge |
||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Goals of care |
Pain management |
Hospice evaluation |
Nonpain symptom management |
Support |
|||||||||||
|
n = 1156 (33.8%) |
n = 430 (12.6%) |
n = 301 (8.8%) |
n = 1397 (40.9%) |
n = 133 (3.9%) |
|||||||||||
| Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | Controls | Consult | p | |
| Variable costs | |||||||||||||||
| 0–1 |
3375 |
2876 |
0.238 |
3272 |
3218 |
0.652 |
2905 |
1487 |
<0.001* |
2929 |
2861 |
0.782 |
2993 |
2725 |
0.915 |
| 2–3 |
4389 |
3839 |
0.41 |
3883 |
4316 |
0.081 |
3714 |
2675 |
<0.001* |
3433 |
4016 |
0.192 |
4229 |
3924 |
0.934 |
| 4–5 |
4435 |
4585 |
0.892 |
4505 |
5733 |
0.027* |
4472 |
3821 |
0.096 |
3925 |
4892 |
0.195 |
4683 |
7463 |
0.815 |
| 6–10 |
6295 |
7576 |
0.496 |
6898 |
7488 |
0.48 |
5914 |
5849 |
0.918 |
6038 |
7604 |
0.21 |
16,916 |
7859 |
0.117 |
| Length of stay | |||||||||||||||
| 0–1 |
6.1 |
5.3 |
<0.001* |
5 |
5.6 |
<0.001* |
5 |
2.7 |
<0.001* |
4.8 |
5.6 |
0.008* |
6.3 |
6.2 |
0.874 |
| 2–3 |
7 |
6.8 |
0.439 |
6.4 |
7.2 |
<0.001* |
6 |
4.2 |
<0.001* |
6.7 |
7.5 |
0.191 |
7 |
7.6 |
0.408 |
| 4–5 |
9.3 |
9 |
0.485 |
8.5 |
9.9 |
0.009* |
7.9 |
6 |
<0.001* |
10.1 |
12.5 |
0.059 |
10.9 |
10.9 |
0.99 |
| 6–10 |
12.8 |
13.3 |
0.322 |
12.8 |
13.4 |
0.481 |
11.3 |
9.3 |
0.001* |
13 |
14.4 |
0.402 |
15.3 |
17.4 |
0.345 |
| Total unit operating costs | |||||||||||||||
| 0–1 |
8844 |
7646 |
<0.001* |
7254 |
8123 |
<0.001* |
7296 |
3817 |
<0.001* |
6197 |
6516 |
0.394 |
6987 |
8397 |
0.091 |
| 2–3 |
10,600 |
10,363 |
0.483 |
10,612 |
12,637 |
<0.001* |
10,020 |
7149 |
<0.001* |
11,088 |
12,167 |
0.242 |
10,575 |
10,626 |
0.956 |
| 4–5 |
13,447 |
12,721 |
0.217 |
13,931 |
16,290 |
0.022* |
12,448 |
9639 |
<0.001* |
12,866 |
15,395 |
0.117 |
15,672 |
19,520 |
0.171 |
| 6–10 | 18,600 | 19,666 | 0.219 | 21,772 | 23,352 | 0.343 | 17,421 | 15,493 | 0.088 | 18,737 | 21,777 | 0.219 | 23,312 | 27,171 | 0.241 |
p < 0.05.
Authors' Contributions
All authors have contributed to the development and production of this article and have approved this draft.
Funding Statement
Dr. Kavalieratos receives funding from NHLBI (K01HL133466), Cystic Fibrosis Foundation (KAVAL19QI0), and Milbank Foundation.
Author Disclosure Statement
No competing financial interests exist.
References
- 1. Kavalieratos D, Corbelli J, Zhang D, et al. : Association between palliative care and patient and caregiver outcomes. JAMA 2016;316:2104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Morrison RS, Dietrich J, Ladwig S, et al. : Palliative care consultation teams cut hospital costs for Medicaid beneficiaries. Health Aff 2011;30:454–463. [DOI] [PubMed] [Google Scholar]
- 3. Bajwah S, Oluyase AO, Yi D, et al. : The effectiveness and cost-effectiveness of hospital-based specialist palliative care for adults with advanced illness and their caregivers. Cochrane Database Syst Rev 2020;2020:CD012780. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Ernecoff NC, Check D, Bannon M, et al. : Comparing specialty and primary palliative care interventions: Analysis of a systematic review. J Palliat Med 2019:jpm.2019.0349. DOI: 10.1089/jpm.2019.0349. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. May P, Garrido MM, Cassel JB, et al. : Prospective cohort study of hospital palliative care teams for inpatients with advanced cancer: Earlier consultation is associated with larger cost-saving effect. J Clin Oncol 2015;33:2745–2752. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ernecoff NC, Wessell KL, Hanson LC, et al. : Elements of palliative care in the last 6 months of life: Frequency, predictors, and timing. J Gen Intern Med 2019. DOI: 10.1007/s11606-019-05349-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Barkley JE, McCall A, Maslow AL, et al. : Timing of palliative care consultation and the impact on thirty-day readmissions and inpatient mortality. J Palliat Med 2019;22:393–399. [DOI] [PubMed] [Google Scholar]
- 8. May P, Garrido MM, Cassel B, et al. : Cost analysis of a prospective multi-site cohort study of palliative care consultation teams for adults with advanced cancer: Where do cost-savings come from? Palliat Med 2017;31:378–386. [DOI] [PubMed] [Google Scholar]
- 9. Quill TE, Abernethy AP: Generalist plus specialist palliative care—creating a more sustainable model. N Engl J Med 2013;368:1173–1175. [DOI] [PubMed] [Google Scholar]
- 10. Kamal AH, Bull JH, Swetz KM, et al. : Future of the palliative care workforce: Preview to an impending crisis. Am J Med 2017;130:113–114. [DOI] [PubMed] [Google Scholar]
- 11. Ferrell BR, Twaddle ML, Melnick A, Meier DE: Special report national consensus project clinical practice guidelines for quality palliative care guidelines, 4th edition. J Palliat Med 2018;21:1684–1689. [DOI] [PubMed] [Google Scholar]
- 12. Kavalieratos D: Directing the narrative to define and present standardization in palliative care. J Palliat Med 2019;22:1486–1487. [DOI] [PubMed] [Google Scholar]
- 13. Quan H, Sundararajan V, Halfon P, et al. : Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005;43:1130–1139. [DOI] [PubMed] [Google Scholar]
- 14. Averill RF, Goldfield N, Hughes JS, et al. All Patient Refined Diagnosis Related Groups (APR-DRGs): Methodology Overview. Wallingford, CT: 3M Health Information Systems; 2003. [Google Scholar]
- 15. Mahalanobis PC: On the generalised distance in statistics. Proc Natl Inst Sci India 1936;2:49–55. [Google Scholar]
- 16. Suissa S: Immortal time bias in pharmacoepidemiology. Am J Epidemiol 2008;167:492–499. [DOI] [PubMed] [Google Scholar]
- 17. Ernecoff NC, Wessell KL, Hanson LC, et al. : Does receipt of recommended elements of palliative care precede in-hospital death or hospice referral? J Pain Symptom Manage 2020;59:778–786. [DOI] [PubMed] [Google Scholar]
- 18. Trotti A, Colevas AD, Basch E, et al. : Patient-reported outcomes and the evolution of adverse event reporting in oncology. Artic J Clin Oncol 2007;25:5121–5127. [DOI] [PubMed] [Google Scholar]
- 19. Kavalieratos D: Reading past the p <0.05's: The secondary messages of systematic reviews and meta-analyses in palliative care. Palliat Med 2019;33:121–122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Wells RR, Dionne-Odom JR, Azuero A, et al. : Examining adherence and dose effect of an early palliative care intervention for advanced heart failure patients. J Pain Sympt Manage 2021; Feb5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Kamal AH. Letters “‘who does what?’” ensuring high-quality and coordinated palliative care with our oncology colleagues. J Pain Symptom Manage 2016;52:e1–e2. [DOI] [PubMed] [Google Scholar]
- 22. Kaufman BG, Van Houtven CH, Greiner MA, et al. : Selection bias in observational studies of palliative care: Lessons learned. J Pain Symptom Manage 2020. DOI: 10.1016/j.jpainsymman.2020.09.011. [DOI] [PubMed] [Google Scholar]
- 23. Greer DS, Mor V, Morris JN, et al. : An alternative in terminal care: Results of the National Hospice Study. J Chronic Dis 1986;39:9–26. [DOI] [PubMed] [Google Scholar]
- 24. Mor V, Kidder D: Cost savings in hospice: Final results of the National Hospice Study. Health Serv Res 1985;20:407–422. [PMC free article] [PubMed] [Google Scholar]