BACKGROUND:

Pediatric somatic symptom and related disorders (SSRDs) are common with high health care costs and use because of lack of standardized, evidence-based practice. Our hospital implemented a clinical pathway (CP) for SSRD evaluation and management. Our study objective was to evaluate health care cost and use associated with the organization’s SSRD CP in the emergency department (ED) and inpatient settings hypothesizing lower cost and use in the CP group relative to controls.

METHODS:

We conducted a retrospective analysis of costs and use before and after implementation of the SSRD CP. Data were collected from the hospital’s electronic health record and the Pediatric Health Information System database. Participants included pediatric patients on the CP (“P” group) and control groups with an SSRD diagnosis and mental health consultation either the year before the CP (“C” group) or during the CP study period (“T” group). Primary outcomes included costs, length of stay, diagnostic testing, imaging, subspecialty consultation, and readmission rates.

RESULTS:

The ED P group had more lower-cost imaging, whereas the inpatient T group greater higher-cost imaging than other groups. The inpatient P group had significantly shorter length of stay, fewer subspecialty consults, and lower costs. There were no significant group differences in readmission rates. The CP reduced median total costs per patient encounter by $51 433 for the inpatient group and$6075 for the ED group.

CONCLUSIONS:

The CP group showed significant reductions in health care cost and use after implementation of a CP for SSRD care. In future work, researchers should explore patient and practitioner experience with the SSRD CP and long-term outcomes.

Somatic symptom and related disorders (SSRDs) are characterized by physical symptoms inconsistent with physical disease, significantly influenced by psychological distress, and result in marked and persistent impairment.1  SSRDs are common in pediatric hospitals. They are associated with limited application of evidence-based practices, wide variation in care, and significant health care cost and use.27  In previous studies, health care cost and length of stay (LOS) for pediatric SSRDs were improved with early mental health consultation and concurrent medical evaluation.8  Rehabilitation, close care coordination, and psychological interventions, informed by a thoughtful multidisciplinary biopsychosocial assessment, are known to reduce symptom burden, disability, and school absence in pediatric patients with SSRDs.9,10

Significant SSRD-related pediatric hospital cost and use results primarily from a lack of standardized care.11  Clinical pathways (CPs) can effectively address challenges in health care delivery for problems with high prevalence, high care variability, and available evidence-based practices.12  CPs can improve several aspects of care including LOS, cost, and complications.13,14  There has been a growing interest in the use of CP in pediatrics to address diseases such as pneumonia, otitis media, and cystic fibrosis yet little use of CPs for the evaluation and management of mental health conditions.11,1517  Although in some studies authors have found that diagnostic and practice guidelines for SSRDs improve clinical outcomes and/or treatment attendance,18,19  no evaluation of cost and use patterns has been published.

At our institution, a multidisciplinary working group of pediatric providers developed and implemented a hospital-wide CP for pediatric SSRDs in October 2015.20  Our local CP, which includes consult orders, communication strategies and timing, and discharge planning, aligns well with a recently published national consensus CP for SSRD evaluation and management11,20  and was informed by local resources and workflows to enhance operationalization. It is described in a separate publication.20  In this study, our purpose was to evaluate changes in health care cost and use associated with this local CP with the following hypotheses:

• 1. Primary hypothesis

• a. This CP would lead to reduced health care costs.

• 2. Secondary hypotheses based on factors thought to be associated with cost reduction

• a. Because LOS and the number of subspecialty consultations can lead to an increase in health care costs, youth with SSRDs admitted to the emergency department (ED) or inpatient service and placed on the CP would see a reduction in both of these variables.

• b. Youth on the CP would receive earlier mental health consultations relative to the control groups.

This study was conducted at a 226-bed tertiary care academic pediatric hospital housing pediatric general, neonatal, and cardiothoracic ICUs, a pediatric ED, an embedded child and adolescent psychiatry hospital, and embedded pediatric physical medicine and rehabilitation service, with access to all approved pediatric subspecialty services. There are independent pediatric psychology and psychiatry consultation services, an active child life service, and a psychiatric ED. This study was granted approval from the institution’s institutional review board.

#### Pediatric Health Information System

Data for this study were obtained from the Pediatric Health Information System (PHIS), an administrative database that contains inpatient, ED, ambulatory surgery, and observation encounter-level data from >50 not-for-profit tertiary care pediatric hospitals in the United States.21  These hospitals are affiliated with the Children’s Hospital Association (Lenexa, KS). Data quality and reliability are assured through a joint effort between the Children’s Hospital Association and participating hospitals. For the purposes of external benchmarking, participating hospitals provide discharge and/or encounter data including demographics, diagnoses, and procedures. Nearly all of these hospitals also submit resource use data (eg, pharmaceuticals, imaging, and laboratory) into PHIS. Data are deidentified at the time of data submission and are subjected to a number of reliability and validity checks before being included in the database. For this study, only data from our hospital were included and consisted of the following categories: ED or inpatient admission, sex, age, race, ethnicity, insurance class, admission source, LOS, and encounter costs. Billed hospital charges were converted to costs by using cost-to-charge ratios and included the total, clinical, pharmacy, laboratory, imaging, supply, and other (ie, room, nursing, operating room, and ED) costs.

#### Electronic Health Record

Additional data were extracted from the patient electronic health record by the data translation office and chart review and include the reason for visit, subspecialist consult, diagnostic testing (electrocardiogram, EEG), lower-cost imaging (radiograph, echocardiogram, ultrasound), higher-cost imaging (MRI, computed tomography, nuclear medicine), procedures (pulmonary function testing, nasogastric tube placement), revisit within 30 days,22,23  and calendar day of mental health consultation.

#### Prepathway Control Group

A prepathway control group (“C” group) included the following:

1. youth seen in the ED for a somatic complaint and discharged with an SSRD diagnosis; or

2. youth admitted with a somatic complaint, diagnosed with SSRD, and received a psychiatry or psychology consult during their admission between September 2014 and September 2015 (Fig 1).

FIGURE 1

Participants. Flowchart of participant inclusion criteria.

FIGURE 1

Participants. Flowchart of participant inclusion criteria.

This 1-year prepathway time frame was limited by the dates of our hospital’s initial participation in the PHIS.

#### Pathway Group

The pathway group (“P” group) included the following:

1. youth placed on the SSRD CP between October 2015 and December 2017; and

2. either discharged from the ED or admitted to the hospital.

An automatic e-mail was sent to the primary investigators when the order was placed.

#### Time-Match Group

Because of concern that not all patients who qualified for the CP would be universally enrolled, a second postpathway time-match group (“T” group) of youth were identified who were not formally placed on the CP and were either as follows:

1. seen in the ED for a somatic complaint and discharged with an SSRD diagnosis; or

2. admitted with a somatic complaint, diagnosed with SSRD, and received a psychiatry or psychology consult during their admission.

The rationale for the T group was to assess differences in health care cost and use related to culture change and education that occurred over time as this CP and its tenets of care became the new norm for care of any youth with suspected SSRD (comparing to the C group) or related to formal placement on the CP with more standardized care approaches (comparing to the P group).

#### Patients Excluded From the Study

Patients were excluded from the study for the following reasons:

1. They had an SSRD diagnosis but their admission was not for a somatic complaint.

2. They were directly admitted to the physical medicine inpatient rehabilitation service.

3. They had a psychiatric ED rather than a pediatric ED visit.

4. They had an SSRD diagnosis of psychological factors affecting medical condition for nonadherence to treatment of a chronic medical condition rather than a somatic complaint.

Analyses were conducted by using SAS 9.4 (SAS Institute, Inc, Cary, NC). Univariate statistics were used to describe the sample and to check if dependent variables were normally distributed. χ2 or Fisher’s exact tests were used to test for group differences for categorical variables. Analysis of variance (ANOVA) or Kruskal-Wallis tests were used to test for group differences in dependent variables with normal and nonnormal distributions, respectively.

Significant intergroup contrasts are shown in Tables 1 through 5.

TABLE 1

Patient Characteristics

CharacteristicsED (n = 71)Inpatient (n = 162)
ControlTime MatchPathwaySignificant ContrastsaP bControlTime MatchPathwaySignificant ContrastsaP b
n = 19n = 20n = 32n = 53n = 54n = 55
Sex, n (%)    C ≠ P .09    — .16
Female 11 (58) 13 (65) 27 (84) — — 42 (79) 37 (69) 46 (84) — —
Male 8 (42) 7 (35) 5 (16) — — 11 (21) 17 (31) 9 (16) — —
Age (SD) 13.6 (2.4) 14.9 (2.2) 13.5 (2.5) T ≠ P .08 13.9 (2.3) 14.0 (2.5) 14.1 (2.7) — .92
Race, n (%)    — .46    — .78
White 13 (68) 15 (75) 27 (84) — — 29 (74) 42 (78) 41 (75) — —
Black/African American 4 (21) 3 (15) 4 (13) — — 10 (19) 11 (20) 10 (18) — —
Asian American 1 (5) 0 (0) 1 (3) — — 1 (2) 1 (2) 2 (4) — —
Other 1 (5) 2 (10) 0 (0) — — 3 (6) 0 (0) 2 (4) — —
Ethnicity, n (%)    T ≠ P .04    — .54
Non-Hispanic 19 (100) 17 (85) 32 (100) — — 52 (98) 51 (94) 51 (93) — —
Hispanic 0 (0) 3 (15) 0 (0) — — 1 (2) 3 (6) 4 (7) — —
Insurance, n (%)    — .62    — .44
Commercial 12 (63) 15 (75) 24 (75) —  32 (60) 33 (61) 39 (71) — —
Government 7 (37) 5 (25) 8 (25) —  21 (40) 21 (39) 16 (29) — —
Reason for admission, n (%)    C, T ≠ P .02    T ≠ P .04
Pain 4 (21) 5 (25) 19 (59) — — 16 (30) 25 (46) 18 (33) — —
Neurology 13 (68) 14 (70) 12 (38) — — 33 (63) 20 (37) 31 (56) — —
GI 0 (0) 0 (0) 0 (0) — — 1 (2) 4 (7) 0 (0) — —
Cardiac 1 (5) 1 (5) 1 (3) — — 2 (4) 3 (6) 6 (11) — —
Other 1 (5) 0 (0) 0 (0) — — 1 (2) 2 (4) 0 (0)
Admission source, n (%)    C ≠ T .01    — .66
Home 19 (100) 14 (70) 29 (91) — — 26 (49) 9 (17) 36 (65) — —
Clinic and/or procedure visit 0 (0) 0 (0) 0 (0) — — 10 (19) 10 (19) 7 (13) — —
Transfer ED or urgent care 0 (0) 6 (30) 3 (9) — — 14 (26) 1 (2) 10 (18) — —
Transfer from other hospital 0 (0) 0 (0) 0 (0) — — 14 (26) 9 (17) 2 (4) — —
CharacteristicsED (n = 71)Inpatient (n = 162)
ControlTime MatchPathwaySignificant ContrastsaP bControlTime MatchPathwaySignificant ContrastsaP b
n = 19n = 20n = 32n = 53n = 54n = 55
Sex, n (%)    C ≠ P .09    — .16
Female 11 (58) 13 (65) 27 (84) — — 42 (79) 37 (69) 46 (84) — —
Male 8 (42) 7 (35) 5 (16) — — 11 (21) 17 (31) 9 (16) — —
Age (SD) 13.6 (2.4) 14.9 (2.2) 13.5 (2.5) T ≠ P .08 13.9 (2.3) 14.0 (2.5) 14.1 (2.7) — .92
Race, n (%)    — .46    — .78
White 13 (68) 15 (75) 27 (84) — — 29 (74) 42 (78) 41 (75) — —
Black/African American 4 (21) 3 (15) 4 (13) — — 10 (19) 11 (20) 10 (18) — —
Asian American 1 (5) 0 (0) 1 (3) — — 1 (2) 1 (2) 2 (4) — —
Other 1 (5) 2 (10) 0 (0) — — 3 (6) 0 (0) 2 (4) — —
Ethnicity, n (%)    T ≠ P .04    — .54
Non-Hispanic 19 (100) 17 (85) 32 (100) — — 52 (98) 51 (94) 51 (93) — —
Hispanic 0 (0) 3 (15) 0 (0) — — 1 (2) 3 (6) 4 (7) — —
Insurance, n (%)    — .62    — .44
Commercial 12 (63) 15 (75) 24 (75) —  32 (60) 33 (61) 39 (71) — —
Government 7 (37) 5 (25) 8 (25) —  21 (40) 21 (39) 16 (29) — —
Reason for admission, n (%)    C, T ≠ P .02    T ≠ P .04
Pain 4 (21) 5 (25) 19 (59) — — 16 (30) 25 (46) 18 (33) — —
Neurology 13 (68) 14 (70) 12 (38) — — 33 (63) 20 (37) 31 (56) — —
GI 0 (0) 0 (0) 0 (0) — — 1 (2) 4 (7) 0 (0) — —
Cardiac 1 (5) 1 (5) 1 (3) — — 2 (4) 3 (6) 6 (11) — —
Other 1 (5) 0 (0) 0 (0) — — 1 (2) 2 (4) 0 (0)
Admission source, n (%)    C ≠ T .01    — .66
Home 19 (100) 14 (70) 29 (91) — — 26 (49) 9 (17) 36 (65) — —
Clinic and/or procedure visit 0 (0) 0 (0) 0 (0) — — 10 (19) 10 (19) 7 (13) — —
Transfer ED or urgent care 0 (0) 6 (30) 3 (9) — — 14 (26) 1 (2) 10 (18) — —
Transfer from other hospital 0 (0) 0 (0) 0 (0) — — 14 (26) 9 (17) 2 (4) — —

C, control; GI, gastrointestinal; P, clinical pathway; T, Time match. —, not applicable.

a

Group differences indicated, P < .05.

b

Reported P for 3-way ANOVA.

TABLE 2

MeasuresEffect Size, Odds Ratio (95% CI)
Control, n (%)Time Match, n (%)Pathway, n (%)Significant ContrastsaPbC Versus TC Versus PT Versus P
n = 19n = 20n = 32
Received nonmental health subspecialty consult 6 (32) 4 (20) 4 (13) — .24 0.54 (0.13–2.34) 0.31 (0.07–1.29) 1.75 (0.38–7.97)
Diagnostic testing 3 (16) 4 (20) 5 (16) — .92 1.33 (0.26–6.94) 0.99 (0.21–4.70) 1.35 (0.32–5.77)
Higher-cost imaging 2 (11) 3 (15) 3 (9) — .89 1.50 (0.22–10.14) 0.88 (0.13–5.80) 1.71 (0.30–9.42)
Lower-cost imaging 3 (16) 2 (10) 14 (44) C, T ≠ P .01 0.59 (0.09–4.01) 4.15 (1.01–17.11) 0.14 (0.03–0.72)
Procedures 0 (0) 0 (0) 0 (0) — — — — —
30-d readmission 1 (5) 2 (10) 3 (9) — .5 2.00 (0.17–24.07) 1.86 (0.18–19.30) 1.07 (0.16–7.06)
MeasuresEffect Size, Odds Ratio (95% CI)
Control, n (%)Time Match, n (%)Pathway, n (%)Significant ContrastsaPbC Versus TC Versus PT Versus P
n = 19n = 20n = 32
Received nonmental health subspecialty consult 6 (32) 4 (20) 4 (13) — .24 0.54 (0.13–2.34) 0.31 (0.07–1.29) 1.75 (0.38–7.97)
Diagnostic testing 3 (16) 4 (20) 5 (16) — .92 1.33 (0.26–6.94) 0.99 (0.21–4.70) 1.35 (0.32–5.77)
Higher-cost imaging 2 (11) 3 (15) 3 (9) — .89 1.50 (0.22–10.14) 0.88 (0.13–5.80) 1.71 (0.30–9.42)
Lower-cost imaging 3 (16) 2 (10) 14 (44) C, T ≠ P .01 0.59 (0.09–4.01) 4.15 (1.01–17.11) 0.14 (0.03–0.72)
Procedures 0 (0) 0 (0) 0 (0) — — — — —
30-d readmission 1 (5) 2 (10) 3 (9) — .5 2.00 (0.17–24.07) 1.86 (0.18–19.30) 1.07 (0.16–7.06)

C, control; N/A, not available; P, clinical pathway; T, time match; —, not applicable.

a

Group differences indicated, P < .05.

b

Reported P for 3-way ANOVA.

TABLE 3

MeasuresGroupsEffect Size, η2 or Odds Ratio (95% CI)
ControlTime MatchPathwaySignificant ContrastsaPbC Versus TC Versus PT Versus P
n = 53n = 54n = 55
LOS, IQR (range) 2, 1–3 (1–16) 2, 1–3 (1–12) 1, 1–2 (1–11) C ≠ P .04 0.01 0.06 0.02
Received nonmental health subspecialty consult, n (%) 30 (57) 21 (39) 13 (24) C ≠ P .002 0.49 (0.22–1.05) 0.24 (0.10–0.54) 2.06 (0.90–4.71)
Diagnostic testing, n (%) 25 (47) 19 (35) 23 (42) — .45 0.61 (0.28–1.32) 0.81 (0.38–1.72) 0.76 (0.35–1.64)
Higher-cost imaging, n (%) 13 (25) 20 (37) 10 (18) T ≠ P .08 1.81 (0.79–4.17) 0.68 (0.27–1.73) 2.65 (1.10–6.38)
Lower-cost imaging, n (%) 16 (30) 21 (39) 12 (22) — .15 1.47 (0.66–3.28) 0.65 (0.27–1.54) 2.28 (0.98–5.29)
Procedures, n (%) 9 (17) 7 (13) 3 (5) — .13 0.73 (0.25–2.12) 0.28 (0.07–1.11) 2.58 (0.63–10.56)
30-d readmission, n (%) 5 (9) 6 (11) 7 (13) — .86 1.20 (0.34–4.20) 1.40 (0.42–4.72) 0.86 (0.27–2.74)
MeasuresGroupsEffect Size, η2 or Odds Ratio (95% CI)
ControlTime MatchPathwaySignificant ContrastsaPbC Versus TC Versus PT Versus P
n = 53n = 54n = 55
LOS, IQR (range) 2, 1–3 (1–16) 2, 1–3 (1–12) 1, 1–2 (1–11) C ≠ P .04 0.01 0.06 0.02
Received nonmental health subspecialty consult, n (%) 30 (57) 21 (39) 13 (24) C ≠ P .002 0.49 (0.22–1.05) 0.24 (0.10–0.54) 2.06 (0.90–4.71)
Diagnostic testing, n (%) 25 (47) 19 (35) 23 (42) — .45 0.61 (0.28–1.32) 0.81 (0.38–1.72) 0.76 (0.35–1.64)
Higher-cost imaging, n (%) 13 (25) 20 (37) 10 (18) T ≠ P .08 1.81 (0.79–4.17) 0.68 (0.27–1.73) 2.65 (1.10–6.38)
Lower-cost imaging, n (%) 16 (30) 21 (39) 12 (22) — .15 1.47 (0.66–3.28) 0.65 (0.27–1.54) 2.28 (0.98–5.29)
Procedures, n (%) 9 (17) 7 (13) 3 (5) — .13 0.73 (0.25–2.12) 0.28 (0.07–1.11) 2.58 (0.63–10.56)
30-d readmission, n (%) 5 (9) 6 (11) 7 (13) — .86 1.20 (0.34–4.20) 1.40 (0.42–4.72) 0.86 (0.27–2.74)

C, control; IQR, interquartile range; P, clinical pathway; T, time match; —, not applicable.

a

Group differences indicated, P < .05.

b

Reported P for 3-way ANOVA.

TABLE 4

Costs: ED

Costs, Median (IQR), $Effect Size, η2 ControlTime MatchPathwaySignificant ContrastsaPbC Versus TC Versus PT Versus P Total 8704(4362–16248) 2029(957–3887) 2629(1287–3976) C ≠ T, P <.0001 0.32 0.31 0.01 Clinical 1236(330–2779) 389(118–681) 561(232–935) C ≠ T, P .007 0.21 0.13 0.02 Pharmacy 5(0–40) 0(0–42) 3(0–72) — .58 0.02 0.00 0.01 Laboratory 1548(0–2665) 59(0–532) 57(0–510) C ≠ T, P .05 0.09 0.11 0.00 Imaging 0(0–0) 0(0–0) 226(0–933) — .07 0.00 0.07 0.06 Supply 0(0–0) 0(0–0) 0(0–0) — .25 0.02 0.03 0.00 Other 6153(4019–6401) 880(845–1347) 1293(863–1377) C ≠ T, P <.0001 0.49 0.45 0.03 Costs, Median (IQR),$Effect Size, η2
ControlTime MatchPathwaySignificant ContrastsaPbC Versus TC Versus PT Versus P
Total 8704(4362–16248) 2029(957–3887) 2629(1287–3976) C ≠ T, P <.0001 0.32 0.31 0.01
Clinical 1236(330–2779) 389(118–681) 561(232–935) C ≠ T, P .007 0.21 0.13 0.02
Pharmacy 5(0–40) 0(0–42) 3(0–72) — .58 0.02 0.00 0.01
Laboratory 1548(0–2665) 59(0–532) 57(0–510) C ≠ T, P .05 0.09 0.11 0.00
Imaging 0(0–0) 0(0–0) 226(0–933) — .07 0.00 0.07 0.06
Supply 0(0–0) 0(0–0) 0(0–0) — .25 0.02 0.03 0.00
Other 6153(4019–6401) 880(845–1347) 1293(863–1377) C ≠ T, P <.0001 0.49 0.45 0.03

C, control; IQR, interquartile range; P, clinical pathway; T, time match; —, not applicable.

a

Group differences indicated, P < .05.

b

Reported P for 3-way ANOVA.

TABLE 5

Costs: Inpatient

Costs, Median (IQR), $Effect Size, η2 ControlTime MatchPathwaySignificant ContrastsaPbC Versus TC Versus PT Versus P Total 60 369 (47 293–95 134) 12 695 (8673–20 159) 8926 (6136–13 307) C ≠ T ≠ P <.0001 0.60 0.65 0.07 Clinical 19 841 (6719–35 996) 3814 (1184–6523) 1514 (888–5686) C ≠ T, P <.0001 0.41 0.49 0.02 Pharmacy 542 (134–1913) 210 (19–777) 80 (7–280) C ≠ T ≠ P <.0001 0.06 0.22 0.06 Laboratory 1811 (35 529–44 2631) 738 (279–1309) 195 (0–839) C ≠ T ≠ P <.0001 0.08 0.19 0.08 Imaging 0 (0–9661) 226 (0–5229) 0 (0–834) C, T ≠ P .07 0.00 0.04 0.04 Supply 298 (0–450) 0 (0–65) 0 (0–60) C ≠ T, P <.0001 0.12 0.16 0.01 Other 24 178 (18 549–42 229) 5212 (3590–7222) 4173 (3532–6222) C ≠ T, P <.0001 0.52 0.57 0.01 Costs, Median (IQR),$Effect Size, η2
ControlTime MatchPathwaySignificant ContrastsaPbC Versus TC Versus PT Versus P
Total 60 369 (47 293–95 134) 12 695 (8673–20 159) 8926 (6136–13 307) C ≠ T ≠ P <.0001 0.60 0.65 0.07
Clinical 19 841 (6719–35 996) 3814 (1184–6523) 1514 (888–5686) C ≠ T, P <.0001 0.41 0.49 0.02
Pharmacy 542 (134–1913) 210 (19–777) 80 (7–280) C ≠ T ≠ P <.0001 0.06 0.22 0.06
Laboratory 1811 (35 529–44 2631) 738 (279–1309) 195 (0–839) C ≠ T ≠ P <.0001 0.08 0.19 0.08
Imaging 0 (0–9661) 226 (0–5229) 0 (0–834) C, T ≠ P .07 0.00 0.04 0.04
Supply 298 (0–450) 0 (0–65) 0 (0–60) C ≠ T, P <.0001 0.12 0.16 0.01
Other 24 178 (18 549–42 229) 5212 (3590–7222) 4173 (3532–6222) C ≠ T, P <.0001 0.52 0.57 0.01

C, control; IQR, interquartile range; P, clinical pathway; T, time match.

a

Group differences indicated, P < .05.

b

Reported P for 3-way ANOVA.

Overall, the study population was predominantly female and white, with ages ranging from 7 to 18, similar to previous studies of pediatric SSRDs in the hospital setting (Table 1).24  There were no statistically significant demographic differences between patients discharged from the ED compared with those admitted.

For the ED groups, the P group had significantly more girls than both control groups, whereas the T group was significantly more Hispanic and older compared with the P group. There were no differences in health insurance status (government versus commercial). The P group was significantly more likely to be seen for pain complaints. The T group was significantly more likely to be transferred from an outside ED or urgent care than the C group.

Inpatient groups were demographically similar. There were no significant differences in health insurance status or admission source. Youth in the T group were significantly more likely to be admitted for pain and less likely to be admitted for neurologic symptoms than the P group.

For ED patients, there were no statistically significant differences in subspecialist consultation rates, diagnostic testing, higher-cost imaging, procedures, or 30-day readmission rates (Table 2). Patients on the CP received more lower-cost imaging than the other groups (P versus C odds ratio 4.15 [95% confidence interval (CI) 1.01–17.11]; P versus T odds ratio 0.14 [95% CI 0.03–0.72]).

Inpatients on the CP had significantly shorter LOS (η2 = 0.06) and fewer subspecialty consults than the C group (odds ratio 0.24 [95% CI 0.10–0.54]). The T group received significantly more higher-cost imaging than the P group (odds ratio 2.65 [95% CI 1.10–6.38]). There were no group differences in diagnostic testing, lower-cost imaging, procedures, and 30-day readmission rates (Table 3). Eighty-five percent of the P group received psychology and 44% received psychiatry consults. Frequencies are not reported for the control groups because psychology and/or psychiatry consultation were part of the inclusion criteria. There were no group differences in the median day of psychology or psychiatry consultation (1 day).

For patients discharged from the ED, median total costs were significantly higher in the C group relative to the P and T groups ($8704 vs$2629 vs $2029; Table 4, Fig 2) resulting in$6075 saved per ED patient encounter. Clinical (η2 T versus C = 0.21; P versus C = 0.13) and other (η2 T versus C = 0.49; P versus C = 0.45) costs were higher in the C group than in the P and T groups. C group laboratory costs were significantly higher than those of the P group (η2 = 0.11). There were no significant group differences in pharmacy, imaging, and supply costs.

FIGURE 2

Costs of ED and inpatient admissions by group. A, Total ED costs (C ≠ T, P; P < .0001). The effect size for C versus T = 0.32 and for C versus P = 0.31. B, Categorical ED costs. For clinical, laboratory, and other, C ≠ T, P (P = .007); .05 (P < .001). The effect sizes are shown in Table 4. C, Total inpatient costs (C ≠ T ≠ P; P < .0001). The effect size for C versus T = 0.60, for C versus P = 0.62, and for T versus P = 0.07. D, Categorical inpatient costs. Three-way ANOVAs are all significant to P < .0001. The effect sizes are shown in Table 5. a Denotes nonsignificant difference.

FIGURE 2

Costs of ED and inpatient admissions by group. A, Total ED costs (C ≠ T, P; P < .0001). The effect size for C versus T = 0.32 and for C versus P = 0.31. B, Categorical ED costs. For clinical, laboratory, and other, C ≠ T, P (P = .007); .05 (P < .001). The effect sizes are shown in Table 4. C, Total inpatient costs (C ≠ T ≠ P; P < .0001). The effect size for C versus T = 0.60, for C versus P = 0.62, and for T versus P = 0.07. D, Categorical inpatient costs. Three-way ANOVAs are all significant to P < .0001. The effect sizes are shown in Table 5. a Denotes nonsignificant difference.

For inpatients, median total costs were significantly different between all groups, with gradated reductions from C to T to P groups ($60 369 vs$12 695 vs $8926; η2 T versus C = 0.60; P versus C = 0.65; T versus P = 0.07) (Table 5, Fig 2) resulting in$51 433 saved per inpatient encounter. Clinical (η2 T versus C = 0.41; P versus C = 0.49), supply (η2 T versus C = 0.12; P versus C = 0.16), and other (η2 T versus C = 0.52; P versus C = 0.57) costs were higher in the C group than in the post-CP implementation groups. There were significant gradated differences across groups for pharmacy (η2 T versus C = 0.06; P versus C = 0.22; T versus P = 0.06) and laboratory (η2 T versus C = 0.08; P versus C = 0.19; T versus P = 0.08) costs. Imaging costs were lower in the P group relative to C and T groups (η2 CP versus C = 0.04; T versus P = 0.04).

A review of risk management reports indicated no reports or adverse events (ie, youth incorrectly diagnosed with SSRD, family reports of harm or distress) for youth on the CP.

This study is the first reported description of health care use and cost in a pediatric hospital associated with development and implementation of an evidence and consensus-based SSRD CP for the ED and inpatient settings.11,20  Our results support the initial hypotheses that a CP for pediatric SSRD care reduces cost and use. The CP did not adversely impact care, with no known youth being inappropriately placed on the CP and compromising the identification of physical disease. Notably, effect sizes for these cost reductions were large for P versus C groups (η2 = 0.31 for ED; 0.65 for inpatient) and small to moderate for P versus T groups (η2 = 0.01 for ED; 0.07 for inpatient). On the basis of conservative estimates of 15 ED and 26 inpatient patient SSRD CP encounters annually, estimated annual cost savings would be $1 428 383 (15 ×$6075 + 26 × $51 433). There are several factors that likely contributed to these cost reductions (Fig 3). First, we found that CP implementation resulted in fewer and more targeted subspecialist consultations with likely downstream additive impacts on cost and use patterns. Second, early mental health consultation, a key component of this CP, has been found to be a factor in LOS reduction and hospitalization costs.8 However, we did not see hypothesized improvement in timing of mental health consultation because we were not anticipating that the C group would have a median day of mental health consult of 1. Third, reduced inpatient LOS likely also contributed to cost reduction. Because high occupancy and timely discharges are significant issues for pediatric hospitals, interventions that reduce patient LOS can allow for improved patient access and reduced provider frustration.25,26 Although this LOS reduction may appear small, rapid return to functioning in youth with SSRD is a key component of recovery and a core focus of intervention.27,28 FIGURE 3 Conceptualization of potential direct and indirect impacts of CP implementation on SSRD care. FIGURE 3 Conceptualization of potential direct and indirect impacts of CP implementation on SSRD care. Although the CP itself offers advantages around standardizing care, enhancing communication, and managing expectations, it is suggested in this study that there may be associated culture changes that appear to impact practice compared to care before the CP. Notably, there were incremental reductions in costs, LOS, and subspecialty consultation along the continuum from the C to T to P groups. This finding may be related to changes in training, education, awareness, culture, and practice associated with dissemination and repeated use of the CP, which may have improved confidence in managing SSRDs.29 Provider education about the CP’s focus on communication strategies, shared expectations for goals of care, and normalization of early mental health consultation may have impacted outcomes regardless of whether an individual patient was placed on the CP.8 Simply providing more awareness of SSRDs as a diagnostic category in the hospital and the differential factors in the care of this population compared with other youth with physical disease may have impacted care practices that were beyond the scope of this study. However, the moderate effect sizes seen when comparing T to P, particularly for inpatient groups, suggests that the CP does improve use and cost more than these factors alone. Despite significant improvements in health care cost and use, the CP did not reduce overall diagnostic medical workup.30 This is highlighted by the lack of significant differences in the number of diagnostic tests, imaging, and procedures between inpatient C and P groups. For youth discharged from the ED, the P group had higher rates of lower-cost imaging, which may have been necessary to appropriately exclude a diagnosis and safely discharge the patient. In addition, the inpatient T group’s higher rates of diagnostic testing relative to the P group may be a clue to why they were not placed on the CP and is suggestive of perhaps a more complicated medical picture, possible familial influences on practice, or lack of provider confidence in the SSRD diagnosis after admission. Together, these findings suggest that implementation of a CP can reduce cost and use without compromising the integrity and fidelity of medical diagnostic evaluation or care. The study outcomes address the concern that CP implementation in SSRD care will limit evaluation of physical health factors, prolong care, or potentially miss physical disease.30 This study reveals the opposite: implementation of the CP in the ED resulted in as much, if not more, diagnostic evaluation by using lower-cost diagnostic tools to ensure completion of proper medical evaluation. Lack of mental health consultation in the ED may have influenced medical workup patterns, although the ability to measure this is beyond the scope of this study. Implementation in the inpatient setting resulted in medical evaluation that was not compromised but more targeted with greater use of lower-cost testing. Our CP has a number of factors impacting reduction in unnecessary workup including multicomponent intervention, targeting both providers and families, and targeting systemic changes.31 The study is retrospective and nonrandomized in design limiting the ability to make causal inferences. As a single-site study designed around our specific CP, the findings may not generalize to other sites that may have variable access to mental health and subspecialty resources or different health system resources. Although we did not find any documentation of adverse events, lack of follow-up of patients diagnosed with SSRDs limited our ability to determine if any harm came from delayed medical diagnosis and should be explored in further research. In addition, the T group is not a true control group because these patients were not randomly assigned. Rather, they were separated because of clinical decision-making, possible failure of providers to enroll the patient in pathway despite eligibility, or other unknown reasons. It is unclear why this occurred and may be due to medical complexity, familial influences, or lack of provider awareness, knowledge, or confidence in standardized SSRD care in our institution. Future investigation should be used to explore decision-making patterns regarding youth placed on the CP and those that are not to better understand the differences between these populations and their care. Furthermore, inclusion in the T and C group required a diagnosis of an SSRD and a mental health consultation, both identified in the electronic health record. Before CP implementation, there was no standardized approach to routinely consulting mental health for this population or documenting the SSRD diagnosis. Therefore, the study did not include youth who did not receive mental health consultation or those who did not receive a formal SSRD diagnosis by their medical provider. Because this population was not captured, it is possible that the impact of the CP on cost and use may be greater than reported, suggested by changes seen in the T group despite nonenrollment in the CP. Study design and analyses were also limited by which data were available for inclusion. Specifically, data could not be accessed regarding the use of outpatient services and services at hospitals not enrolled in PHIS including other regional children’s hospitals. Additionally, although no direct funding was used, professional time investments were not recorded or calculated as costs of implementation. Quantification of downstream effects on provider, nursing, and staff experience and productivity were also beyond the scope of this study. Thus, a true cost-effectiveness analysis could not be completed. Furthermore, the study setting has unique access to psychiatric and psychological consultation services in the inpatient setting but not the ED. The lack of mental health consultation in the ED could significantly impact patterns of communication, diagnostic workup, and disposition because joint mental and physical health evaluation and management is a cardinal feature of SSRD care. The CP, through its directed implementation of standardized evidence-based practices, may explain the improvements in health care costs and use in the ED, but it is unclear how mental health consultation in the ED may further impact outcomes. This is a subject for future study. Finally, in this study, we did not control for outside factors like provider education and experience that may have impacted SSRD practice patterns. Although it is suggested in other studies that SSRD training, education, and standardized practice in medical settings are generally sparse,29 this was not directly measured. As a result, it is not known if these variables may have impacted which youth were or were not enrolled in the CP, leading to differences between T and P groups. Furthermore, in this study, we were not able to sufficiently capture which aspects of CP implementation were more routinely adopted into practice, resulting in a potential consideration for future study. Using the same data set, we plan to examine institutional disparities in SSRD care and CP impacts on these disparities by race and ethnicity, sex, insurance status, and socioeconomic status, as well as history of other mental health diagnoses. We will investigate provider perceptions of the CP and SSRDs and possible institutional culture changes surrounding SSRDs using qualitative research methods with inpatient pediatric providers and resident physicians. Future research could be used to investigate whether similar CPs can lead to similar outcome improvements at other institutions and whether CPs can be useful for other common mental health conditions in youth cared for in pediatric medical inpatient and ED settings. Finally, determining patient and family satisfaction and clinical outcomes with SSRD CPs will also be important going forward. This study reveals that a CP for pediatric SSRDs can address the high costs and health care use challenges associated with care of this population. This SSRD CP led to large cost reductions, with estimated annual cost savings of$1 428 383. Furthermore, this CP did not receive grant or other direct funding allocation beyond professionals’ use of time for meetings and CP development, suggesting the implementation of the CP requires limited upfront costs and is sustainable over time with access to appropriate mental health consultation resources. The exercise of developing a CP for pediatric SSRDs was highly valuable at our institution. Our results suggest that locally specific CPs, such as the CP described in this study, can be developed and adapted broadly and can be reasonably expected to reduce cost and use of inpatient and ED care for pediatric SSRDs.

We thank our institution’s SSRD working group for their contribution in the development of the SSRD CP that we describe here. We thank Heather Crossley for providing the PHIS data and Amrutha Kurakula, MPH, for her contribution to the early phases of data management and conceptualization.

Dr Kullgren conceptualized and designed the study, secured funding, and drafted the initial manuscript; Mr Klein conducted data analyses and drafted the initial manuscript; Dr Malas conceptualized and designed the study and drafted the initial manuscript; Drs Monroe, Sroufe, and Hutton conceptualized and designed the study; Ms Sturza conducted data analyses and drafted parts of the manuscript; Ms Pardon collected data; and all authors reviewed and revised the manuscript and approved the final manuscript as submitted.

FUNDING: Funded by the University of Michigan Department of Pediatrics Children’s Health Research Award.

1
American Psychiatric Association
.
Diagnostic and Statistical Manual of Mental Disorders (DSM-5).
5th ed.
Arlington, VA
:
American Psychiatric Association Publishing
;
2013
2
Bardach
NS
,
Coker
TR
,
Zima
BT
, et al
.
Common and costly hospitalizations for pediatric mental health disorders
.
Pediatrics
.
2014
;
133
(
4
):
602
609
3
Doupnik
SK
,
Lawlor
J
,
Zima
BT
, et al
.
Mental health conditions and medical and surgical hospital utilization
.
Pediatrics
.
2016
;
138
(
6
):
e20162416
4
Græsholt-Knudsen
T
,
Skovgaard
AM
,
Jensen
JS
,
CU
.
Impact of functional somatic symptoms on 5-7-year-olds’ healthcare use and costs
.
Arch Dis Child
.
2017
;
102
(
7
):
617
623
5
van der Leeuw
G
,
Gerrits
MJ
,
Terluin
B
, et al
.
The association between somatization and disability in primary care patients
.
J Psychosom Res
.
2015
;
79
(
2
):
117
122
6
Zima
BT
,
Rodean
J
,
Hall
M
,
Bardach
NS
,
Coker
TR
,
Berry
JG
.
Psychiatric disorders and trends in resource use in pediatric hospitals
.
Pediatrics
.
2016
;
138
(
5
):
e20160909
7
Barsky
AJ
,
Orav
EJ
,
Bates
DW
.
Somatization increases medical utilization and costs independent of psychiatric and medical comorbidity
.
Arch Gen Psychiatry
.
2005
;
62
(
8
):
903
910
8
Bujoreanu
S
,
White
MT
,
Gerber
B
,
Ibeziako
P
.
Effect of timing of psychiatry consultation on length of pediatric hospitalization and hospital charges
.
Hosp Pediatr
.
2015
;
5
(
5
):
269
275
9
Bonvanie
IJ
,
Kallesøe
KH
,
Janssens
KAM
,
Schröder
A
,
Rosmalen
JGM
,
CU
.
Psychological interventions for children with functional somatic symptoms: a systematic review and meta-analysis
.
J Pediatr
.
2017
;
187
:
272.e17
281.e17
10
Aiarzaguena
JM
,
Gaminde
I
,
Clemente
I
,
Garrido
E
.
Explaining medically unexplained symptoms: somatizing patients’ responses in primary care
.
Patient Educ Couns
.
2013
;
93
(
1
):
63
72
11
Ibeziako
P
,
Brahmbhatt
K
,
Chapman
A
, et al
.
Developing a clinical pathway for somatic symptom and related disorders in pediatric hospital settings
.
Hosp Pediatr
.
2019
;
9
(
3
):
147
155
12
Lawal
AK
,
Rotter
T
,
Kinsman
L
, et al
.
What is a clinical pathway? Refinement of an operational definition to identify clinical pathway studies for a Cochrane systematic review
.
BMC Med
.
2016
;
14
:
35
13
Rotter
T
,
Kinsman
L
,
James
E
, et al
.
Clinical pathways: effects on professional practice, patient outcomes, length of stay and hospital costs
.
Cochrane Database Syst Rev
.
2010
;(
3
):
CD006632
14
Seehusen
DA
.
Clinical pathways: effects on practice, outcomes, and costs
.
Am Fam Physician
.
2010
;
82
(
11
):
1338
1339
15
Dona
D
,
Baraldi
M
,
G
, et al
.
The impact of clinical pathways on antibiotic prescribing for acute otitis media and pharyngitis in the emergency department
.
Pediatr Infect Dis J
.
2018
;
37
(
9
):
901
907
16
Kaiser
SV
,
Rodean
J
,
Bekmezian
A
, et al
.
Effectiveness of pediatric asthma pathways for hospitalized children: a multicenter, national analysis
.
J Pediatr
.
2018
;
197
:
165.e2
171.e2
17
Singh
SB
,
Shelton
AU
,
Greenberg
B
,
Starner
TD
.
Implementation of cystic fibrosis clinical pathways improved physician adherence to care guidelines
.
Pediatr Pulmonol
.
2017
;
52
(
2
):
175
181
18
Drane
DL
,
LaRoche
SM
,
Ganesh
GA
,
Teagarden
D
,
Loring
DW
.
A standardized diagnostic approach and ongoing feedback improves outcome in psychogenic nonepileptic seizures
.
Epilepsy Behav
.
2016
;
54
:
34
39
19
Löwe
B
,
Piontek
K
,
Daubmann
A
, et al
.
Effectiveness of a stepped, collaborative, and coordinated health care network for somatoform disorders (Sofu-Net): a controlled cluster cohort study
.
Psychosom Med
.
2017
;
79
(
9
):
1016
1024
20
Kullgren
KA
,
Shefler
A
,
Malas
N
, et al
.
Taking the pain out of somatization: development and implementation of a hospital-based clinical practice guideline to address pediatric somatic symptom and related disorders
.
Hosp Pediatr
.
2020
;
10
(
2
):
105
113
21
Children’s Hospital Association
.
PHIS
.
2019
.
Available at: https://www.childrenshospitals.org/phis. Accessed October 22, 2019
22
Auger
KA
,
Teufel
RJ
II
,
Harris
JM
II
, et al
.
Children’s hospital characteristics and readmission metrics
.
Pediatrics
.
2017
;
139
(
2
):
e20161720
23
Axon
RN
,
Williams
MV
.
Hospital readmission as an accountability measure
.
JAMA
.
2011
;
305
(
5
):
504
505
24
Bujoreanu
S
,
Randall
E
,
Thomson
K
,
Ibeziako
P
.
Characteristics of medically hospitalized pediatric patients with somatoform diagnoses
.
Hosp Pediatr
.
2014
;
4
(
5
):
283
290
25
Fieldston
ES
,
Hall
M
,
Sills
MR
, et al
.
Children’s hospitals do not acutely respond to high occupancy
.
Pediatrics
.
2010
;
125
(
5
):
974
981
26
Stockwell
DC
,
Thomas
C
,
Fieldston
ES
, et al
.
Using length of stay to understand patient flow for pediatric inpatients
.
Pediatr Qual Saf
.
2017
;
3
(
1
):
e050
27
Campo
JV
,
Comer
DM
,
Jansen-Mcwilliams
L
,
Gardner
W
,
Kelleher
KJ
.
Recurrent pain, emotional distress, and health service use in childhood
.
J Pediatr
.
2002
;
141
(
1
):
76
83
28
Malas
N
,
Ortiz-Aguayo
R
,
Giles
L
,
Ibeziako
P
.
Pediatric somatic symptom disorders
.
Curr Psychiatry Rep
.
2017
;
19
(
2
):
11
29
Malas
N
,
Donohue
L
,
Cook
RJ
,
Leber
SM
,
Kullgren
KA
.
Pediatric somatic symptom and related disorders: primary care provider perspectives
.
Clin Pediatr (Phila)
.
2018
;
57
(
4
):
377
388
30
Doupnik
SK
,
Dunn
M
.
A novel pathway for somatic symptoms: strategies for successful pathway implementation
.
Hosp Pediatr
.
2019
;
9
(
3
):
223
224
31
Hiscock
H
,
Neely
RJ
,
Warren
H
,
Soon
J
,
Georgiou
A
.
Reducing unnecessary imaging and pathology tests: a systematic review
.
Pediatrics
.
2018
;
141
(
2
):
e20172862

## Competing Interests

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

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.