OBJECTIVES

To characterize patterns of surgery among pediatric patients during terminal hospitalizations in children’s hospitals.

METHODS

We reviewed patients ≤20 years of age who died among 4 424 886 hospitalizations from January 2013–December 2019 within 49 US children’s hospitals in the Pediatric Health Information System database. Surgical procedures, identified by International Classification of Diseases procedure codes, were classified by type and purpose. Descriptive statistics characterized procedures, and hypothesis testing determined if undergoing surgery varied by patient age, race and ethnicity, or the presence of chronic complex conditions (CCCs).

RESULTS

Among 33 693 terminal hospitalizations, the majority (n = 30 440, 90.3%) of children were admitted for nontraumatic causes. Of these children, 15 142 (49.7%) underwent surgery during the hospitalization, with the percentage declining over time (P < .001). When surgical procedures were classified according to likely purpose, the most common were to insert or address hardware or catheters (31%), explore or aid in diagnosis (14%), attempt to rescue patient from mortality (13%), or obtain a biopsy (13%). Specific CCC types were associated with undergoing surgery. Surgery during terminal hospitalization was less likely among Hispanic children (47.8%; P < .001), increasingly less likely as patient age increased, and more so for Black, Asian American, and Hispanic patients compared with white patients (P < .001).

CONCLUSIONS

Nearly half of children undergo surgery during their terminal hospitalization, and accordingly, pediatric surgical care is an important aspect of end-of-life care in hospital settings. Differences observed across race and ethnicity categories of patients may reflect different preferences for and access to nonhospital-based palliative, hospice, and end-of-life care.

What’s Known on This Subject:

The majority of infants, children, and adolescents who die in the United States do so in a hospital. Although 15% of adult patients experiencing a terminal hospitalization undergo surgery, little is known about surgical procedures during pediatric terminal hospitalizations.

What This Study Adds:

Nearly half of terminal hospitalizations involved surgical procedures. Surgery was less common among Hispanic patients and became less common as patients aged, an effect more pronounced among multiracial patients. Pediatric surgical care is an important aspect of goal-directed end-of-life care.

Each year, ∼45 000 infants, children, and adolescents in the United States die.1  Although some of these deaths occur suddenly with no opportunity for medical care, the majority occur after some period of medical care, typically in a hospital setting.2  Surgery plays an important yet still underappreciated role in palliative and end-of-life care.3,4  Pediatric surgical interventions provided near the end of life may intend to attempt cure or life prolongation or to address palliative goals such as symptom management (eg, pain or symptom reduction) or facilitation of supportive care (eg, placement of gastrostomy tubes or provision of vascular access).5,6  These range of surgical goals have been described for pediatric patients with various life-threatening conditions, such as advanced heart disease,7,8  short bowel syndrome,9  and late-stage cancer.6 

Previous researchers have described the use of surgery at the end-of-life among elderly patients,10  with the percentage of patients undergoing surgery ranging from 16.8% during terminal hospitalization11  to 25% in the last month of life12  and 31.9% in the last year of life.10  In the adult realm, research priorities identified for advancing palliative care in surgery have included measuring outcomes of importance to patients and families, ways to improve communication and shared decision-making, and how to effectively integrate and deliver palliative care to surgical patients.4  Although such research priorities also pertain to pediatric care, only 1 single-institution study has examined surgical procedures among children at the end of life.13  Furthermore, although unexpected pediatric deaths can occur during or shortly after surgery, these are much rarer events than in the adult realm, and pediatric surgery during a terminal hospitalization is more likely to be pursuing a goal of attempting rescue, clarifying the clinical predicament, or providing palliation. A greater understanding of pediatric surgical intervention at the end of life would help inform efforts to improve goal-concordant care.

Therefore, in this national-level multicenter case series of pediatric terminal hospitalizations, we describe the incidence, type, and likely purpose of surgical procedures performed near or at the end-of-life for hospitalized infants, children, and adolescents. Given the known differences in mortality across the pediatric age1416  range, medical care needs based on underlying conditions,1720  and end-of-life care patterns by race and ethnicity,2124  we also examined variation in surgical care by age, underlying chronic complex conditions (CCCs),25  and race and ethnicity.

We performed a retrospective case series study of children who died during hospital admission from January 2013 to December 2019 within 49 US children’s hospitals using the Pediatric Health Information System (Children's Hospital Association, Lenexa, KS). Pediatric Health Information System is an administrative and billing database of hospital encounters that includes up to 41 diagnoses and procedure codes for each encounter by using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and 10th Revision, Clinical Modification (ICD-10-CM) system. The International Classification of Diseases procedure codes not only include surgical procedures but also include nonsurgical procedures such as spinal taps and computed tomography scans. In the study, we used deidentified administrative data that is considered nonhuman subjects research by the institutional review board at the Children’s Hospital of Philadelphia.

We included all patients (age ≤20 years; the upper age limit for many hospitals) who died during hospitalization within the study period; we further identified those who underwent a surgical procedure. Surgical patients were identified by using ICD-9-CM and ICD-10-CM procedural codes. We defined surgery broadly during our initial data extraction: the presence of any ICD-9-CM or ICD-10-CM procedure code. Patients admitted for trauma were identified via the Agency for Healthcare Research and Quality’s Clinical Classification Software categories 2601 to 2620; because nearly all hospitalized trauma patients who die undergo surgical interventions attempting rescue before their death, they were excluded from the analysis.

All ICD-9-CM and ICD-10-CM codes were reviewed and narrowed to include only surgical procedures. Authors M.D.T. and M.N.C. independently reviewed and classified 1290 ICD-9-CM and 6751 ICD-10-CM codes. Authors M.D.T., M.N.C., and R.M.A. reached a consensus on classification of surgical procedures. Surgical procedures were classified by body systems involved and likely purpose. Body systems were adapted from the Accreditation Council for Graduate Medical Education Residency Review Committee for Surgery26  and chosen with attention to systems served by surgical subspecialties. Each code was classified into a body system category based on the primary operative site.

We identified 7 likely surgical purposes: insertion or adjustments of hardware or catheters, attempts to rescue patient from mortality, exploration, biopsy, resection of tissue, care for cardiac congenital conditions, and care for noncardiac congenital conditions. Attempts to rescue patient from mortality included procedures that included efforts to control bleeding or organ transplant. Biopsies were specifically mentioned in ICD-9-CM and ICD-10-CM codes, whereas only resection of tissue was described in other codes. Procedures describing exploration, suturing, repair, and drainage were classified as exploratory or aiding in diagnosis. The body system and likely purpose classification system and associated ICD-9-CM and ICD-10-CM codes used in this case series is included in the Supplemental Information.

Prematurity was identified by diagnostic codes (code range for ICD-9-CM from 765.21 to 765.28; ICD-10-CM, P07.21 to P07.39). Each code specified an estimated gestational age. To represent age across the entire cohort, each premature infant was assigned an age in days before term birth.

CCCs, categorized into groups of related underlying conditions, were identified on the basis of ICD-9 CM and ICD-10 CM codes according to the published classification system for CCC version 2.25  The CCC classification system has been used extensively in previous studies of pediatric complex care1720  and end-of-life medical care.2729 

Our main analytic objective was to describe the patterns of surgical care in this large multicenter case series of patients during terminal hospitalizations. For categorical variables (such as exposure to a surgical procedure, or a type of procedure), we calculated percentages. For continuous variables (such as patient age) we calculated medians with interquartile ranges (IQRs). We compared demographic and clinical characteristics between terminal hospitalizations of patients with and without exposure to surgical procedures using χ2 and Wilcoxon rank tests. We tested for associations between the exposure to a surgical procedure and patient age (at time of death for all tests of associations), race, and types of CCCs using logistic regression. All statistical analyses were performed by using Stata 16.1 (Stata corp, College Station, TX) and accounted for clustering of observations by hospitals, with statistical significance set at P < .05.

Over the study period, 33 693 (0.76%) of 4 424 886 patients admitted across 49 children’s hospitals died during hospitalization (Table 1). Among the terminal hospitalizations, the majority 30 440 (90.3%) of patients were admitted for nontraumatic causes. Among these 30 440 patients, the median duration of hospitalization was 7 days (IQR: 2–28). Overall, 15 142 (49.7%) nontrauma patients in terminal hospitalizations underwent a procedure classified as surgical. Hispanic children were less likely to undergo surgery during terminal hospitalization: 47.0% of Hispanic children underwent surgery compared with 51.9% of white children (P < .001).

TABLE 1

Demographic Characteristics of patients Experiencing Terminal Hospitalization, Including Final Analytic Sample of Nontrauma Patients Who Underwent Surgery

OverallNo SurgeryaUnderwent Surgery
Any SurgeryaTraumaFinal Sample, Nontrauma
No. patients 33 693 17 171 16 522 1380 15 142 
Age at death: Median [IQR], y 0.4 [0–6.4] 0.3 [0–6.4] 0.5 [0.1–6.4] 3.8 [0.8–11.9] 0.4 [0.1–5.4] 
 Birth to 27 d, n (%) 10 888 (32.3) 6787 (39.5) 4101 (24.8) 55 (4.0) 4046 (26.7) 
 28 d to 12 mo, n (%) 9114 (27.1) 3300 (19.2) 5814 (35.2) 339 (24.6) 5475 (36.2) 
 13–24 mo, n (%) 1730 (5.1) 868 (5.1) 862 (5.2) 114 (8.3) 748 (4.9) 
 2–5 y, n (%) 3310 (9.8) 1801 (10.5) 1509 (9.1) 287 (20.8) 1222 (8.1) 
 6–11 y, n (%) 3270 (9.7) 1726 (10.1) 1544 (9.3) 245 (17.8) 1299 (8.6) 
 12–18 y, n (%) 4740 (14.1) 2381 (13.9) 2359 (14.3) 319 (23.1) 2040 (13.5) 
19 y and older, n (%) 641 (1.9) 308 (1.8) 333 (2.0) 21 (1.5) 312 (2.1) 
Premature birth status,bn (%) 9674 (28.7) 5021 (29.2) 4653 (28.2) 87 (6.3) 4566 (30.2) 
Sex, n (%)      
 Male 18 602 (55.4) 9452 (55.3) 9150 (55.5) 795 (58.2) 8355 (55.3) 
 Female 14 982 (44.3) 7646 (44.3) 7336 (44.2) 571 (41.8) 6765 (44.7) 
Race and ethnicity, n (%)      
 Non-Hispanic white 14 068 (41.8) 6887 (40.1) 7181 (43.5) 630 (45.7) 6551 (43.3) 
 Non-Hispanic Black 6331 (18.8) 3287 (19.1) 3044 (18.4) 299 (21.7) 2745 (18.1) 
 Hispanic 6118 (18.2) 3260 (19.0) 2858 (17.3) 217 (15.7) 2641 (17.4) 
 Asian American 1156 (3.4) 568 (3.3) 588 (3.6) 26 (1.9) 562 (3.7) 
 Other 6020 (17.9) 3169 (18.5) 2851 (17.3) 208 (15.1) 2643 (17.5) 
Payer, n (%)      
 Government 19 880 (59) 10 139 (59) 9741 (59) 831 (60.2) 8910 (58.8) 
 Private 12 470 (37) 6405 (37.3) 6065 (36.7) 473 (34.3) 5592 (36.9) 
 Other 1343 (4) 627 (3.7) 716 (4.3) 76 (5.5) 640 (4.2) 
OverallNo SurgeryaUnderwent Surgery
Any SurgeryaTraumaFinal Sample, Nontrauma
No. patients 33 693 17 171 16 522 1380 15 142 
Age at death: Median [IQR], y 0.4 [0–6.4] 0.3 [0–6.4] 0.5 [0.1–6.4] 3.8 [0.8–11.9] 0.4 [0.1–5.4] 
 Birth to 27 d, n (%) 10 888 (32.3) 6787 (39.5) 4101 (24.8) 55 (4.0) 4046 (26.7) 
 28 d to 12 mo, n (%) 9114 (27.1) 3300 (19.2) 5814 (35.2) 339 (24.6) 5475 (36.2) 
 13–24 mo, n (%) 1730 (5.1) 868 (5.1) 862 (5.2) 114 (8.3) 748 (4.9) 
 2–5 y, n (%) 3310 (9.8) 1801 (10.5) 1509 (9.1) 287 (20.8) 1222 (8.1) 
 6–11 y, n (%) 3270 (9.7) 1726 (10.1) 1544 (9.3) 245 (17.8) 1299 (8.6) 
 12–18 y, n (%) 4740 (14.1) 2381 (13.9) 2359 (14.3) 319 (23.1) 2040 (13.5) 
19 y and older, n (%) 641 (1.9) 308 (1.8) 333 (2.0) 21 (1.5) 312 (2.1) 
Premature birth status,bn (%) 9674 (28.7) 5021 (29.2) 4653 (28.2) 87 (6.3) 4566 (30.2) 
Sex, n (%)      
 Male 18 602 (55.4) 9452 (55.3) 9150 (55.5) 795 (58.2) 8355 (55.3) 
 Female 14 982 (44.3) 7646 (44.3) 7336 (44.2) 571 (41.8) 6765 (44.7) 
Race and ethnicity, n (%)      
 Non-Hispanic white 14 068 (41.8) 6887 (40.1) 7181 (43.5) 630 (45.7) 6551 (43.3) 
 Non-Hispanic Black 6331 (18.8) 3287 (19.1) 3044 (18.4) 299 (21.7) 2745 (18.1) 
 Hispanic 6118 (18.2) 3260 (19.0) 2858 (17.3) 217 (15.7) 2641 (17.4) 
 Asian American 1156 (3.4) 568 (3.3) 588 (3.6) 26 (1.9) 562 (3.7) 
 Other 6020 (17.9) 3169 (18.5) 2851 (17.3) 208 (15.1) 2643 (17.5) 
Payer, n (%)      
 Government 19 880 (59) 10 139 (59) 9741 (59) 831 (60.2) 8910 (58.8) 
 Private 12 470 (37) 6405 (37.3) 6065 (36.7) 473 (34.3) 5592 (36.9) 
 Other 1343 (4) 627 (3.7) 716 (4.3) 76 (5.5) 640 (4.2) 
a

Comparison of patient characteristics of “no surgery” group to “any surgery” group: age, P < .001 sex, P = .68; race and ethnicity, P < .001; payer, P = .005.

b

Premature birth status assessed for all patients <1 y of age. See methods section for assessment method.

Over time (Fig 1), the percentage of patients undergoing surgery decreased from a maximum of 54% in 2015 to a minimum of 46% in 2019 (trend, P < .001). Similar downward trends were observed stratifying by whether patients were premature infants or not, and whether the surgery was limited to implanting or addressing hardware or not.

FIGURE 1

Percentage of terminal hospitalization patients who underwent surgery over a 7-year time span.

FIGURE 1

Percentage of terminal hospitalization patients who underwent surgery over a 7-year time span.

Close modal

Among patients who underwent surgery (Table 2), the median number of surgical procedures was 3 (range: 1, 41; IQR: 1–6). Nearly half (48%) underwent surgery on the first or second day of the hospitalization, and 19% underwent surgery on the last or next to last day. Across the broad range of terminal hospitalization lengths of stay (Supplemental Fig 5), the last surgical intervention occurred relatively close to the last day.

TABLE 2

Characteristics of Terminal Hospitalizations, Including the Number and Timing of Surgical Procedures, Stratified by Underlying Conditions

OverallNo SurgeryaUnderwent Surgery
Any SurgeryaTraumaFinal Sample,Nontrauma
No. patients 33 693 17 171 16 522 1380 15 142 
Hospital days: median (IQR) 7 (2–25) 3 (1–9) 18 (4–51) 4 (2–17) 20 (5–54) 
ICU days: median (IQR) 1 (0–7) 1 (0–3) 3 (0–20) 3 (1–9) 3 (0–21) 
No. surgical procedures 74838 74838 4647 70191 
No. surgical procedures per patient: median (IQR) 0 (0–2) 3 (1–6) 2 (1–4) 3 (1–6) 
 For patient with neonatal CCCs 1 (0–2) NA 2 (1–5) 3 (1–6) 2 (1–5) 
 For patient with cardiac CCCs 1 (0–5) NA 4 (2–8) 3 (1–6) 4 (2–8) 
 For patient with neurologic CCCs 0 (0–2) NA 2 (1–5) 2 (1–3) 2 (1–5) 
 For patient with metabolic CCCs 1 (0–3) NA 3 (1–6) 2 (1–3) 3 (1–6) 
 For patient with malignancy CCCs 1 (0–3) NA 3 (1–5) 3 (1–6) 3 (1–5) 
Hospital day of first surgery: median (IQR) 0 (0–0) NA 0 (0–0) 0 (0–0) 0 (0–1) 
 For patient with neonatal CCCs 0 (0–0) NA 0 (0–0) 0 (0–0) 0 (0–0) 
 For patient with cardiac CCCs 0 (0–0) NA 0 (0–1) 0 (0–0) 0 (0–1) 
 For patient with neurologic CCCs 0 (0–0) NA 0 (0–0) 0 (0–0) 0 (0–0) 
 For patient with metabolic CCCs 0 (0–1) NA 0 (0–1) 0 (0–0) 0 (0–1) 
 For patient with malignancy CCCs 0 (0–1) NA 0 (0–2) 0 (0–3) 0 (0–2) 
Hospital day of last surgery: median (IQR) 8 (1–31) NA 8 (1–31) 1 (0–6) 9 (1–34) 
 For patient with neonatal CCCs 9 (1–35) NA 9 (1–35) 17 (3–47) 9 (1–35) 
 For patient with cardiac CCCs 15 (2–46) NA 15 (2–46) 2 (0–20) 16 (3–48) 
 For patient with neurologic CCCs 5 (1–26) NA 5 (1–26) 1 (0–3) 7 (1–31) 
 For patient with metabolic CCCs 12 (2–40) NA 12 (2–40) 1 (0–5) 15 (3–44) 
 For patient with malignancy CCCs 16 (3–45) NA 16 (3–45) 21 (6–49) 16 (3–45) 
OverallNo SurgeryaUnderwent Surgery
Any SurgeryaTraumaFinal Sample,Nontrauma
No. patients 33 693 17 171 16 522 1380 15 142 
Hospital days: median (IQR) 7 (2–25) 3 (1–9) 18 (4–51) 4 (2–17) 20 (5–54) 
ICU days: median (IQR) 1 (0–7) 1 (0–3) 3 (0–20) 3 (1–9) 3 (0–21) 
No. surgical procedures 74838 74838 4647 70191 
No. surgical procedures per patient: median (IQR) 0 (0–2) 3 (1–6) 2 (1–4) 3 (1–6) 
 For patient with neonatal CCCs 1 (0–2) NA 2 (1–5) 3 (1–6) 2 (1–5) 
 For patient with cardiac CCCs 1 (0–5) NA 4 (2–8) 3 (1–6) 4 (2–8) 
 For patient with neurologic CCCs 0 (0–2) NA 2 (1–5) 2 (1–3) 2 (1–5) 
 For patient with metabolic CCCs 1 (0–3) NA 3 (1–6) 2 (1–3) 3 (1–6) 
 For patient with malignancy CCCs 1 (0–3) NA 3 (1–5) 3 (1–6) 3 (1–5) 
Hospital day of first surgery: median (IQR) 0 (0–0) NA 0 (0–0) 0 (0–0) 0 (0–1) 
 For patient with neonatal CCCs 0 (0–0) NA 0 (0–0) 0 (0–0) 0 (0–0) 
 For patient with cardiac CCCs 0 (0–0) NA 0 (0–1) 0 (0–0) 0 (0–1) 
 For patient with neurologic CCCs 0 (0–0) NA 0 (0–0) 0 (0–0) 0 (0–0) 
 For patient with metabolic CCCs 0 (0–1) NA 0 (0–1) 0 (0–0) 0 (0–1) 
 For patient with malignancy CCCs 0 (0–1) NA 0 (0–2) 0 (0–3) 0 (0–2) 
Hospital day of last surgery: median (IQR) 8 (1–31) NA 8 (1–31) 1 (0–6) 9 (1–34) 
 For patient with neonatal CCCs 9 (1–35) NA 9 (1–35) 17 (3–47) 9 (1–35) 
 For patient with cardiac CCCs 15 (2–46) NA 15 (2–46) 2 (0–20) 16 (3–48) 
 For patient with neurologic CCCs 5 (1–26) NA 5 (1–26) 1 (0–3) 7 (1–31) 
 For patient with metabolic CCCs 12 (2–40) NA 12 (2–40) 1 (0–5) 15 (3–44) 
 For patient with malignancy CCCs 16 (3–45) NA 16 (3–45) 21 (6–49) 16 (3–45) 

NA, not applicable.

a

Comparison of patient characteristics of “no surgery” group to “any surgery” group: hospital days, P < .01; ICU days, P < .01.

In total, the patients in this study underwent 70 191 surgical procedures. The most commonly coded surgeries (Supplemental Table 4) were chest tube placement (n= 5952, 8%), extracorporeal membrane oxygenation cannulation (n = 5319, 8%), diagnostic bronchoscopy (n = 3127, 4%), percutaneous abdominal drainage (n = 2116, 3%), and ductus arteriosus closure/restriction (n = 1564, 2%).

Surgical procedures (n = 70 191) were cross classified by body system and by likely purpose (Supplemental Table 4). Most surgical procedures involved thoracic (n = 19 713, 28%), gastrointestinal (n = 16 058, 23%), cardiac (n = 13 779, 20%), or vascular and lymphatic (n = 11 444, 16%) systems. Classifications of likely purpose of surgical procedures included insertion or replacement of hardware or catheters (n = 21 881, 31%), exploration (n = 9609, 14%), attempt to rescue patient from mortality (n = 8893, 13%), obtain a biopsy (n = 8792, 13%), and address congenital cardiac conditions (n = 8017, 11%).

Patient age at time of death by itself was not significantly related to the probability of undergoing surgery at the end of life (P = .85). To illustrate the lack of an association, the percentage of patients <1 year of age who underwent surgery was 50% (9640 of 19 178), compared with the percentage of patients 20 years of age was 48% (119 of 224) (P = .49). Notably, however, comparing neonates who died before a month of age (many of whom were born prematurely) to older patients, 38% (4249 of 11 102) of the neonates underwent surgery compared with 56% (10 893 of 19 338) of the older patients (P < .001) (Fig 2, top panel).

FIGURE 2

Percentage of terminal hospitalization patients who underwent a surgical procedure by age at time of death, overall.

FIGURE 2

Percentage of terminal hospitalization patients who underwent a surgical procedure by age at time of death, overall.

Close modal

Patient age was associated with the surgical procedure’s likely purpose (Fig 2, bottom panel). More precisely, whereas age was not associated with the most common procedures: placing or addressing hardware and catheters (odds ratio [OR]: 1.00 [95% confidence interval (CI): 0.99–1.01]), age was associated with the other likely purposes for surgical procedures (Fig 3, top panel). An increase of 1 year in age was associated with decreasing odds of surgery to resect tissue (OR: 0.98 [95% CI: 0.97–0.99]), rescue patients from mortality (OR: 0.98 [95% CI: 0.97–0.99]), explore or aid in diagnosis (OR: 0.94 [95% CI: 0.94–0.95]), and address congenital cardiac (OR: 0.92 [95% CI: 0.90–0.93]) and noncardiac (OR: 0.92 [95% CI: 0.91–0.94]) conditions (all P < .001). Conversely, increasing age was associated with greater odds of surgery to obtain a biopsy specimen (OR: 1.08 [95% CI: 1.07–1.09], P < .001).

FIGURE 3

Among terminal hospitalization patients across the span of ages at the time of hospitalization, percentage who underwent surgical procedures of different likely purposes and by CCC type. aAge span ranges from premature birth estimated gestational age (EGA) on to term birth, then days, months, and years after birth. bNeonatal CCC-associated terminal admissions were present in newborns, whether born at term or prematurely; thereafter, cases were too infrequent to assess the proportion that underwent surgery.

FIGURE 3

Among terminal hospitalization patients across the span of ages at the time of hospitalization, percentage who underwent surgical procedures of different likely purposes and by CCC type. aAge span ranges from premature birth estimated gestational age (EGA) on to term birth, then days, months, and years after birth. bNeonatal CCC-associated terminal admissions were present in newborns, whether born at term or prematurely; thereafter, cases were too infrequent to assess the proportion that underwent surgery.

Close modal

CCCs were present in 96% (n = 29 258) of terminal hospitalization patients, with 51% (n = 14 781) of these patients having undergone surgery during that terminal hospitalization. Across the age range, the specific types of CCCs were associated with the likelihood of undergoing surgery (P < .001 for the interaction between patient age and the types of CCCs) (Fig 3, bottom panel). Adjusting for CCCs, the probability of surgery during terminal hospitalization decreased slightly with each increasing year of age (OR: 0.99 [95% CI: 0.98–0.99]).

In unadjusted analyses, patient race and ethnicity was significantly associated with likelihood of having undergone surgery (P < .001), with the percentages ranging from 51.9% for white patients to 47.0% for Hispanic patients. In multivariable analysis (adjusting for age at death, payer, and CCCs), the probability of undergoing surgery for white patients was 51.7% (95% CI: 49.8–53.5%), for Asian American patients was 50.2% (95% CI: 46.6–53.7%), for Black patients was 49.4% (95% CI: 46.2–52.7%), and for Hispanic patients was 45.6% (95% CI: 42.9–48.3; this is significantly lower than white patients). Furthermore, a notable interaction was evident between patient race and ethnicity and age on the decline in the likelihood of surgery (Fig 4), whereby the decline was greater for Asian American, Black, and Hispanic patients than for white patients.

FIGURE 4

Adjusted probability of surgery declines with patients’ age at time of death and does so more for Black, Hispanic, and Asian American patients. Probability adjusted for patients’ underlying complex chronic conditions and payer. Race and ethnicity category of “other” was omitted; this category had the least decline over the age span, with a probability just >0.5.

FIGURE 4

Adjusted probability of surgery declines with patients’ age at time of death and does so more for Black, Hispanic, and Asian American patients. Probability adjusted for patients’ underlying complex chronic conditions and payer. Race and ethnicity category of “other” was omitted; this category had the least decline over the age span, with a probability just >0.5.

Close modal

In this large case series of terminal hospitalization in 49 children’s hospitals over a 7-year period, nearly half of the 30 440 patients underwent a surgical procedure, and this percentage had declined over time. Most of the patients who underwent surgery experienced multiple surgical procedures. Children with CCCs were more likely to undergo surgery, and the presence of CCCs influenced the likelihood of surgery with age. Hispanic patients were less likely to undergo surgery, and in analyses adjusting for CCCs, the probability of surgery with increasing age decreased more for Black, Hispanic, and Asian American patients, compared with white patients.

Our data suggest an important difference and similarity between the adult and pediatric contexts regarding surgery during terminal hospitalization. Regarding the difference, surgery is more common during terminal hospitalizations of children than adults, although the precise definition of surgery has not been reported in previous work.11  In our study, we found that nearly 1 in 2 children will undergo surgical procedures at the end of life, compared with 1 in 3 adults estimated to have surgery within the last year of life.10 

Regarding the similarity, the likelihood of undergoing surgery decreases at the extremes of very young and very old age. Among adults, rates of surgery during terminal hospitalizations decline once patients are >80 years of age,10  but with evident practice variation,30  potentially because of lack of consensus regarding how to evaluate the benefits and risks of surgical intervention in old age for specific patients.31  Some researchers have proposed better palliative care education for clinicians and better palliative care access for surgery patients as a means to improve quality of life,32  and decrease unnecessary resource use.33,34  Similarly, we found that premature and neonatal patients were less likely to undergo surgery compared with older patients. Researchers in previous studies have found that some clinicians are less willing to perform surgical procedures on neonates in part because of concern about long-term benefits.9  Many clinicians have concerns about long-term outcomes of surgeries performed on infants,35  suggesting a role for palliative care education and access. For any infant, child, or adolescent with a life-threating illness, the potential benefits and risks of surgical procedures must be carefully discussed with the parents in the context of the overall goals of care for their child,3,5  regardless of whether surgeries are done for palliative purposes or to prevent mortality.36  In such circumstances, a multidisciplinary team approach can improve the decision-making process by offering differing knowledge and perspectives.5,37 

In this study, 53% of patients with CCCs underwent surgery, with the percentage varying depending on CCC type. This is consistent with previous work,38  which documented that 36% of children with life-threatening CCCs (a more restrictive definition) underwent surgery during terminal hospital admission, and variation across type of CCC. During what ultimately proved to be terminal hospitalization, patients with CCCs may undergo surgery with hope of improved duration of survival or quality of life.3,6  Our additional finding that the association of surgery with patient age was modified by CCC type suggests that clinicians and surgeons caring for these types of patients used certain surgical procedures at certain ages. Although CCCs have less predictable fatal outcomes than certain adult conditions (such as specific cancers22 ), their chronic persistence and often-known time course of progression allows more temporally or developmentally targeted use of surgical interventions: surgeons know at what age patients are most likely to safely tolerate and benefit from certain surgical procedures.

Importantly, the disparities we observed across the recorded race and ethnicity categories of patients, with Hispanic patients of any age being less likely to undergo surgery, and a much steeper decline in the likelihood of surgical interventions among multiracial patients as they age, warrant further investigation. Although these disparities could be due to bias in hospital-based clinical decision-making, more likely explanations are rooted in differential access across social, economic, and cultural strata in US society to home-based palliative and hospice care options, and differences in preferences for such care, resulting in more hospital-based end-of-life care.23  Such mechanisms have been put forward to explain lower rates of home death among multiracial patients with deaths attributed to CCCs,21  more hospital deaths among multiracial patients with CCCs,22  and higher rates of resuscitation-related deaths during terminal hospitalizations among Black patients.24 

Lastly, we observed a statistically significant decline over this 7-year period in the percentage of terminal hospitalization patients who underwent surgery during the hospitalization. Whether this trend will continue, and what caused the decline, are uncertain. Possible explanations, such as reduction in surgery during terminal hospitalizations because of shifts in goals of care (perhaps due to enhanced use of primary and specialty pediatric palliative care), or due to greater reluctance to perform surgery in high-risk patients, or improved surgical survival rates, should be investigated.

The findings of this study must be interpreted with 5 important limitations kept in mind. First, we have no data regarding, and thus cannot comment on, the necessity or reasonableness of the surgical care delivered. We know that children are much less likely to undergo surgery with a do-not-attempt-resuscitation order39 ; however, patients’ do-not-attempt-resuscitation status was not reliably available in the data source. As mentioned above, ethically sound clinical care of patients with serious illness, seeking cure, enhanced longevity, or greater comfort, often involves surgical interventions40. Second, the level of detail provided by ICD-9-CM and ICD-10-CM codes limits our description of the procedures performed. Although most of the codes categorized as surgery in this analysis were likely to be performed by surgeons, many procedures, such as percutaneous abdominal drainage and chest tube placement, could be performed by other specialties depending on the practice pattern at a given children’s hospital. Third, this study only included patients who died during terminal hospitalizations and did not include (for example) patients who were discharged and died receiving hospice care. Surgeons may have a lower threshold to operate on hospitalized patients who have no plans for subsequent hospice home care compared with those who do have such plans. Fourth, and relatedly, but in a converse manner, we did not identify surgical procedures that might have occurred during a nonterminal hospitalization or as an outpatient procedure in the months to days before death, some of which might have had the goal of cure or disease amelioration, or might have been employed as part of a palliative effort to allow children to reside and ultimately die at home.2,21,41 Finally, although we categorized a likely typology for each surgical procedure, we were not able to assess the actual intention of a specific surgeon for any specific procedure performed on an individual patient.

Pediatric surgical interventions frequently occur during terminal hospitalizations. Although we cannot ascertain whether the underlying goals guiding the care of these patients were focused on cure, life prolongation, comfort, or quality of life enhancement, these findings underscore the importance of surgical care as an aspect of palliative care for hospitalized infants, children, and adolescents.

FUNDING: No external funding.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2021-053611.

Drs Traynor and Antiel conceptualized and designed the study, performed data analysis, drafted the initial manuscript, and reviewed and revised the manuscript; Ms Camazine assisted in data analysis and reviewed and revised the manuscript; Drs Hall and Feudtner conceptualized and designed the study, coordinated and performed data acquisition and management, supervised and performed data analysis, and reviewed and revised the manuscript; Drs Blinman, Nance, Eghtesady, and Lam contributed to the conceptualization of the study and reviewed and revised the manuscript; all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

CCC

chronic complex condition

CI

confidence interval

ICD-10

International Classification of Diseases, 10th Revision

ICD-9

International Classification of Diseases, Ninth Revision

IQR

interquartile range

OR

odds ratio

1
Centers for Disease Control and Prevention
.
Mortality Data
.
Hyattsville, MD
:
National Vital Statistics System
;
2019
2
Feudtner
C
,
DiGiuseppe
DL
,
Neff
JM
.
Hospital care for children and young adults in the last year of life: a population-based study
.
BMC Med
.
2003
;
1
:
3
3
Feudtner
C
,
Blinman
TA
.
The pediatric surgeon and palliative care
.
Semin Pediatr Surg
.
2013
;
22
(
3
):
154
160
4
Lilley
EJ
,
Cooper
Z
,
Schwarze
ML
,
Mosenthal
AC
.
Palliative care in surgery: defining the research priorities
.
Ann Surg
.
2018
;
267
(
1
):
66
72
5
Shelton
J
,
Jackson
GP
.
Palliative care and pediatric surgery
.
Surg Clin North Am
.
2011
;
91
(
2
):
419
428
,
ix
6
Inserra
A
,
Narciso
A
,
Paolantonio
G
,
Messina
R
,
Crocoli
A
.
Palliative care and pediatric surgical oncology
.
Semin Pediatr Surg
.
2016
;
25
(
5
):
323
332
7
Morell
E
,
Thompson
J
,
Rajagopal
S
,
Blume
ED
,
May
R
.
Congenital cardiothoracic surgeons and palliative care: a national survey study
.
J Palliat Care
.
2021
;
36
(
1
):
17
21
8
Krishnamurthy
G
,
Ratner
V
,
Bacha
E
.
Neonatal cardiac care, a perspective
.
Semin Thorac Cardiovasc Surg Pediatr Card Surg Annu
.
2013
;
16
(
1
):
21
31
9
Pet
GC
,
McAdams
RM
,
Melzer
L
, et al
.
Attitudes surrounding the management of neonates with severe necrotizing enterocolitis
.
J Pediatr
.
2018
;
199
:
186
193.e3
10
Kwok
AC
,
Semel
ME
,
Lipsitz
SR
, et al
.
The intensity and variation of surgical care at the end of life: a retrospective cohort study
.
Lancet
.
2011
;
378
(
9800
):
1408
1413
11
Just
E
,
Casarett
DJ
,
Asch
DA
,
Dai
D
,
Feudtner
C
.
Differences in terminal hospitalization care between U.S. men and women
.
J Pain Symptom Manage
.
2016
;
52
(
2
):
205
211
12
Kwok
AC
,
Hu
YY
,
Dodgion
CM
, et al
.
Invasive procedures in the elderly after stage IV cancer diagnosis
.
J Surg Res
.
2015
;
193
(
2
):
754
763
13
Goudreault
M
,
Humbert
N
,
Gauvin
F
, et al
.
Interventions in the operating room for children near end of life: a multidisciplinary approach
.
J Pediatr Surg
.
2018
;
53
(
5
):
1065
1068
14
National Center for Health Statistics, Centers for Disease Control and Prevention
.
Child health
.
2019
.
15
National Center for Health Statistics, Centers for Disease Control and Prevention
.
Infant health
.
2019
.
Available at: https://www.cdc.gov/nchs/fastats/infant-health.htm. Accessed November 16, 2020
16
National Center for Health Statistics, Centers for Disease Control and Prevention
.
Adolescent health
.
2019
.
Available at: https://www.cdc.gov/nchs/fastats/adolescent-health.htm. Accessed November 16, 2020
17
Berry
JG
,
Hall
M
,
Cohen
E
,
O’Neill
M
,
Feudtner
C
.
Ways to identify children with medical complexity and the importance of why
.
J Pediatr
.
2015
;
167
(
2
):
229
237
18
Berry
JG
,
Hall
M
,
Dumas
H
, et al
.
Pediatric hospital discharges to home health and postacute facility care: a national study
.
JAMA Pediatr
.
2016
;
170
(
4
):
326
333
19
Berry
JG
,
Hall
M
,
Neff
J
, et al
.
Children with medical complexity and Medicaid: spending and cost savings
.
Health Aff (Millwood)
.
2014
;
33
(
12
):
2199
2206
20
Simon
TD
,
Berry
J
,
Feudtner
C
, et al
.
Children with complex chronic conditions in inpatient hospital settings in the United States
.
Pediatrics
.
2010
;
126
(
4
):
647
655
21
Feudtner
C
,
Feinstein
JA
,
Satchell
M
,
Zhao
H
,
Kang
TI
.
Shifting place of death among children with complex chronic conditions in the United States, 1989–2003
.
JAMA
.
2007
;
297
(
24
):
2725
2732
22
Johnston
EE
,
Bogetz
J
,
Saynina
O
,
Chamberlain
LJ
,
Bhatia
S
,
Sanders
L
.
Disparities in inpatient intensity of end-of-life care for complex chronic conditions
.
Pediatrics
.
2019
;
143
(
5
):
e20182228
23
Linton
JM
,
Feudtner
C
.
What accounts for differences or disparities in pediatric palliative and end-of-life care? A systematic review focusing on possible multilevel mechanisms
.
Pediatrics
.
2008
;
122
(
3
):
574
582
24
Trowbridge
A
,
Walter
JK
,
McConathey
E
,
Morrison
W
,
Feudtner
C
.
Modes of death within a children’s hospital
.
Pediatrics
.
2018
;
142
(
4
):
e20174182
25
Feudtner
C
,
Feinstein
JA
,
Zhong
W
,
Hall
M
,
Dai
D
.
Pediatric complex chronic conditions classification system version 2: updated for ICD-10 and complex medical technology dependence and transplantation
.
BMC Pediatr
.
2014
;
14
:
199
26
Accreditation Council for Graduate Medical Education
.
Surgery
.
Available at: https://acgme.org/Specialties/Overview/pfcatid/24. Accessed April 13, 2020
27
Lindley
LC
,
Lyon
ME
.
A profile of children with complex chronic conditions at end of life among Medicaid beneficiaries: implications for health care reform
.
J Palliat Med
.
2013
;
16
(
11
):
1388
1393
28
Håkanson
C
,
Öhlén
J
,
Kreicbergs
U
, et al
.
Place of death of children with complex chronic conditions: cross-national study of 11 countries
.
Eur J Pediatr
.
2017
;
176
(
3
):
327
335
29
DeCourcey
DD
,
Silverman
M
,
Oladunjoye
A
,
Balkin
EM
,
Wolfe
J
.
Patterns of care at the end of life for children and young adults with life-threatening complex chronic conditions
.
J Pediatr
.
2018
;
193
:
196
203.e2
30
Schwarze
ML
,
Barnato
AE
,
Rathouz
PJ
, et al
.
Development of a list of high-risk operations for patients 65 years and older
.
JAMA Surg
.
2015
;
150
(
4
):
325
331
31
Yefimova
M
,
Aslakson
RA
,
Yang
L
, et al
.
Palliative care and end-of-life outcomes following high-risk surgery
.
JAMA Surg
.
2020
;
155
(
2
):
138
146
32
Hua
M
,
Lu
Y
,
Ma
X
,
Morrison
RS
,
Li
G
,
Wunsch
H
.
Association between the implementation of hospital-based palliative care and use of intensive care during terminal hospitalizations
.
JAMA Netw Open
.
2020
;
3
(
1
):
e1918675
33
Willmott
L
,
White
B
,
Gallois
C
, et al
.
Reasons doctors provide futile treatment at the end of life: a qualitative study
.
J Med Ethics
.
2016
;
42
(
8
):
496
503
34
Kon
AA
,
Prsa
M
,
Rohlicek
CV
.
Choices doctors would make if their infant had hypoplastic left heart syndrome: comparison of survey data from 1999 and 2007
.
Pediatr Cardiol
.
2013
;
34
(
2
):
348
353
35
Robins
B
,
Booser
A
,
Lantos
JD
.
When parents have misunderstandings about the risks and benefits of palliative surgery
.
Pediatrics
.
2018
;
142
(
6
):
e20180482
36
Balkin
EM
,
Kirkpatrick
JN
,
Kaufman
B
, et al
.
Pediatric cardiology provider attitudes about palliative care: a multicenter survey study
.
Pediatr Cardiol
.
2017
;
38
(
7
):
1324
1331
37
Ananth
P
,
Melvin
P
,
Feudtner
C
,
Wolfe
J
,
Berry
JG
.
Hospital use in the last year of life for children with life-threatening complex chronic conditions
.
Pediatrics
.
2015
;
136
(
5
):
938
946
38
Baumann
LM
,
Williams
K
,
Abdullah
F
,
Hendrickson
RJ
,
Oyetunji
TA
.
Do-not-resuscitate orders and high-risk pediatric surgery: professional nuisance or medical necessity?
J Surg Res
.
2017
;
217
:
213
216
39
Chavoshi
N
,
Miller
T
,
Siden
H
.
Resource utilization among individuals dying of pediatric life-threatening diseases
.
J Palliat Med
.
2013
;
16
(
10
):
1210
1214

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have no financial disclosures relevant to this article to disclose.

Supplementary data