Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Oct 1.
Published in final edited form as: World Neurosurg. 2020 Jul 16;142:e458–e473. doi: 10.1016/j.wneu.2020.07.048

Effect of Comorbid Depression on Surgical Outcomes Following Craniotomy for Malignant Brain Tumors: A Nationwide Readmission Database Analysis

Kavelin Rumalla 1, Michelle Lin 2, Elliot Orloff 2, Li Ding 2, Gabriel Zada 2, William Mack 2, Frank Attenello 2
PMCID: PMC7529986  NIHMSID: NIHMS1613240  PMID: 32682998

Abstract

Objective:

Quality of life is paramount in advanced cancer patients and may be adversely affected by comorbid depression. We hypothesized that comorbid depression is associated with higher rates of postoperative complications, worse functional outcomes, and higher rates of readmission following craniotomy for resection of malignant intracranial tumors.

Methods:

The National Readmissions Database (NRD) was queried from 2010–2014 to identify adult patients undergoing craniotomy for malignant brain tumor resection. The primary psychiatric disease assessed was major depressive disorder (MDD). Outcomes evaluated included neurological or other major complications, incidence of nonroutine discharge, and unplanned readmission at 30- and 90- days.

Results:

Of the 57,621 craniotomies for malignant neoplasms included in the analysis, 11.32% of patients had clinically diagnosed MDD. The presence of MDD was associated with nonroutine discharge (OR 1.10–125, p< 0.0001) as well as higher rates of neurological complications (OR 1.03 – 1.18, p = 0.003). No association between MDD and 30- or 90- day readmissions was noted. Patients with major APR DRG severity and index length of stay greater than 7 days experienced higher rates of 30- and 90- day readmissions.

Conclusions:

There is a clinically significant rate of comorbid MDD in patients with malignant intracranial tumors, and MDD is associated with worse perioperative outcomes. Given the wealth of behavioral and pharmaceutical therapies available, MDD is a modifiable risk factor in this cohort that clinicians should be vigilant in screening for and initiating appropriate treatment protocols.

Keywords: depression, major depressive disorder, malignant brain tumor, cancer, glioma, nationwide, outcomes

INTRODUCTION

Patients with cancer are vulnerable to psychiatric comorbidities, including depression. Clinical depression is estimated to affect ~13–22% of cancer patients, compared to ~5% of the general population.1–3 . However, cancer patients with depression are less likely to be diagnosed and treated, with only 27% receiving adequate care.1

Primary malignant brain tumors commonly arise sporadically and inflict a substantial degree of emotional and physical suffering in patients and families alike. Glioblastoma (grade IV astrocytoma, GBM) is the most common and aggressive of these tumors.4 Aggressive combination therapy is the standard of care, including surgical resection, adjunctive radiotherapy, and chemotherapy. Despite maximal therapy, 5-year survival remains below 5%.5 Therefore, treatment aims directed towards maximize quality of life through prolongation of functional and cognitive status are imperative.

Depression is prevalent in up to 38% of patients with malignant brain tumors, which is amongst the highest of all forms of cancer. 6–8 We hypothesize that comorbid depression is a key modifiable risk factor associated with neurosurgical outcomes in malignant brain tumor patients. In this study, we explored this relationship from a novel perspective, leveraging the strengths of the Nationwide Readmissions Database (NRD), the largest all-payer inpatient database in U.S., which contains data from approximately ~36 million weighted discharges/year. The NRD contains patient linkage variables enable longitudinal tracking across multiple hospitalizations, rendering it superior to the commonly used Nationwide Inpatient Sample (NIS).

We hypothesized that comorbid depression was independently associated with a higher likelihood of post-operative complications, adverse discharge disposition, and 30- and 90-day readmission. This is the first nationally representative study of depression in patients undergoing surgery for malignant brain tumors.

METHODS

Data source

The Nationwide Readmissions Database (NRD) encompasses 27 states which comprise approximately 57.8% of the United States population. Utilizing anonymized identifiers, the NRD tracks patients within a given state over a calendar year. The 2010 to 2014 NRD cohorts were queried for this study.

Study population

Inclusion criteria consisted of patients > 18 years of age, diagnosed with a malignant brain tumor (ICD-9 codes 191.0–191.9, 198.3), who underwent craniotomy for tumor excision or ablation (1.53, 1.59). Patients with above criteria were identified using the International Classification of Diseases, Ninth Edition, Clinical Modification (ICD-9CM) codes of diagnoses and procedures (Table 1). While the NRD allows longitudinal patient tracking, this is limited to a calendar year, with resetting of patient identification each calendar year. Therefore, patients with insufficient follow-up time were excluded from our analysis. The final cohorts only included patients discharged between January and November for the 30-day readmission analysis and patients discharged between January and September for the 90-day readmission analysis. Patients were also excluded from analysis if they died within index hospitalization or were missing key data such as length of stay (LOS).

Table 1.

International Classification of Diseases, Ninth Edition, Clinical Modification Diagnosis Codes and Procedural Codes used for querying patients from the Nationwide Readmission Database for the cohort selection of the present study.

ICD-9 Diagnosis Code Procedure Code
Malignant Intracranial Tumor 191.0–191.9 01.53, 01.59
198.3 1.59
192.0, 192.8, 192.9, 198.3, 198.4 1.53
Psychiatric Diagnosis
Major Depression 296.2x, 296.3x, 296.82, 300.4, 309.0, 309.1, 311
Anxiety 300.0x, 300.2x, 300.3, 309.21, 309.24
Schizophrenia, Psychosis 295.xx, 297.x, 298.x
Bipolar Disorder 296.0x, 296.1x, 296.4x, 296.5x, 296.6x, 296.7, 296.8, 296.80, 296.81, 296.89
Post Traumatic Stress Disorder, Acute Stress Disorder 300.12, 300.13, 300.14, 300.15, 300.6, 308.x, 309.81
Personality Disorders 301.xx
Conversion, Hypochondriasis, Psychogenic Conditions 300.10, 300.11, 300.16, 300.19, 300.5, 300.7, 300.8x, 307.8, 307.80, 307.89, 309.82
Adjustment Disorders 309, 309.2, 309.22, 309.23,309.28, 309.29, 309.3, 309.4, 309.8, 309.83, 309.89, 309.9
Eating Disorders 307.1, 307.5x
Alcohol or Drug Abuse Disorder 303.xx
Tobacco Use 305.1, V15.82
Major Complications
Pneumonia 481–482, 482.1–482.3,482.30–482.32 ,482.39–482.41, 482.49, 482.80–482.84, 482.89, 482.90,483.0 483.1, 483.8, 485–487.0, 997.3 507.0
Pulmonary Embolism 415.1–415.9
Renal Failure 584, 584.5–584.9
Cerebrovascular Accident 433.01, 433.11, 433.21, 433.31, 433.81, 433.91
Myocardial Infarction, Cardiac Arrest 410.00–410.90, 410.01, 410.11– 410.91, 427.5
Sepsis, Septic Shock 995.91, 995.92
Neurological Complications
Intracerebral Hemorrhage 431, 998.11–12
Seizures 345.xx
Neurological Complications After Procedure 997.01–997.09
Open in a new tab

Patient and hospital characteristics

Patients and the index admission hospital variables were obtained from the NRD and analyzed using univariate and multivariate analyses. Hospital variables included procedural volume, ownership (government nonfederal, private nonprofit, private investor-owned), bed size (small, medium, large), and teaching status. Hospitals were categorized as high procedural volume hospitals if they were above the 90th percentile for annual tumor excisions. Patient variables from initial hospitalization included age, gender, payer status, household income quartile based on residence zip code, and whether they resided in the index hospital’s state. The NRD lists age as a continuous variable, which we then categorized into the proceeding groups for analysis: ≤ 40, 45–59, 60–74, and >75 years of age, based on prior categorizations used in NRD studies.

The primary clinical variable assessed was major depression defined by ICD-9 codes 296.2x, 296.3x, 296.82, 300.4, 309.0, 309.1, and 311. The presence of any other psychiatric disease was grouped as a single variable, including anxiety (300.0x, 300.2x, 300.3, 309.21, 309.24), schizophrenia and other psychosis (295.xx, 297.x, 298.x), bipolar disorder (296.0x, 296.1x, 296.4x, 296.5x, 296.6x, 296.7, 296.8, 296.80, 296.81, 296.89), PTSD and acute stress disorders (300.12, 300.13, 300.14, 300.15, 300.6, 308.x, 309.81), personality disorders (301.xx), conversion, hypochondriasis, psychogenic conditions (300.10, 300.11, 300.16, 300.19, 300.5, 300.7, 300.8x, 307.8, 307.80, 307.89, 309.82), miscellaneous adjustment disorders (309, 309.2, 309.22, 309.23,309.28, 309.29, 309.3, 309.4, 309.8, 309.83, 309.89, 309.9), and eating disorders (307.1, 307.5x). Other substance risk factors included alcohol or drug abuse disorder (303.xx), and tobacco use (305.1, V15.82).

General clinical variables included LOS, discharge quarter, Elixhauser comorbidity score, and discharge to a facility other than home. Elixhauser comorbidity score was measured using the Agency for healthcare Research and Quality’s (AHRQ) software, which implements the Elixhauser comorbidity index using ICD-9 codes. Other variables included index admissions were categorized as elective vs nonelective or occurred through an emergency department. All patient refined disease related groups (APR-DRG) risk of disease severity is a risk classification scores for patients (minor, moderate, major, or extreme) used by multiple hospital systems, showing strong association with outcomes.9

Outcome variables queried included index admission occurrence of major complication, neurological complication, and nonroutine discharge, as well as unplanned readmission at 30 and 90 days. Of note, for outcomes of 30- and 90- day unplanned readmission, variables included in models included index admission incidence of major complications, neurological complications, and nonroutine discharge.

Statistical Analysis

Descriptive analysis was reported for all variables mentioned above. Multivariable logistic regression adjusting for depression was used for all outcomes, using generalized estimation equation to account for hospital clustering. All variables mentioned above were used in model initially, with variables not acting as confounders with p > 0.05 were excluded from the final models. All final models passed Hosmer-Lemeshow goodness of fit test. SAS 9.4 were used for all analysis.

RESULTS

Patient and hospital baseline characteristics

We identified 57,621 craniotomies for malignant neoplasms from 2010–2014 that met inclusion criteria for primary admission analysis. There was a comparable distribution of males (53%, n=30,601) to females (47%, n=27,020). The majority of patients were between the ages of 45 to 74 years old (68%, n=39,047) and most had one or more Elixhauser defined comorbidities (78%). 6,525 (11%) patients held a previous diagnosis of depression. Within the group of patients categorized with depression, 31% had a concurrent psychiatric disorder. A predominance of cases came from larger hospitals (79%) that were designated as teaching facilities (80%). The average length of stay was 9.2 days, following which most patients were discharged home (routine disposition, 61%). Patient and hospital baseline characteristics are further summarized in Table 2.

Table 2.

Patient demographics of the 30-day readmission cohort.

Total Depression
No Yes
N N % N %
Total 57621 51096 88.68 6525 11.32
Died during hospitalization Missing 19 16 0.03 DS DS
No 56550 50114 98.08 6436 98.64
Yes 1052 966 1.89 86 1.32
Major complication No 54847 48650 95.21 6197 94.97
Yes 2774 2446 4.79 328 5.03
Neuro complication No 44960 39997 78.28 4963 76.06
Yes 12661 11099 21.72 1562 23.94
Other psychiatric disease No 50449 45916 89.86 4533 69.47
Yes 7172 5180 10.14 1992 30.53
Alcohol and drug abuse disorder No 56942 50554 98.94 6388 97.9
Yes 679 542 1.06 137 2.1
Smoking No 37508 33715 65.98 3793 58.13
Yes 20113 17381 34.02 2732 41.87
Age ≤ 44 12038 11092 21.71 946 14.5
45–59 17969 15621 30.57 2348 35.98
60–74 21078 18457 36.12 2621 40.17
>75 6536 5926 11.6 610 9.35
Sex Male 30601 28019 54.84 2582 39.57
Female 27020 23077 45.16 3943 60.43
Insurance Missing 164 151 0.3 13 0.2
Medicare 20387 17835 34.9 2552 39.11
Medicaid 7193 6422 12.57 771 11.82
Private insurance 25986 23198 45.4 2788 42.73
Self-pay 1646 1488 2.91 158 2.42
No charge 162 147 0.29 15 0.23
Other 2083 1855 3.63 228 3.49
All Patient Refined DRG: Risk of Mortality Subclass No class DS DS DS DS DS
Minor 17728 15946 31.21 1782 27.31
Moderate 19275 17038 33.35 2237 34.28
Major 13770 12059 23.6 1711 26.22
Extreme 6844 6050 11.84 794 12.17
All Patient Refined DRG: Severity of Illness Subclass No class DS DS DS DS DS
Minor 10836 9867 19.31 969 14.85
Moderate 23687 20999 41.1 2688 41.2
Major 17010 14863 29.09 2147 32.9
Extreme 6084 5364 10.5 720 11.03
Lenth of stay ≤ 3 days 17144 15658 30.64 1486 22.77
4–6 days 13804 12418 24.3 1386 21.24
7–11 days 13506 11993 23.47 1513 23.19
> 12 days 13167 11027 21.58 2140 32.8
Discharge quarter Jan-March 14542 12895 25.24 1647 25.24
Apr-Jun 14699 13062 25.56 1637 25.09
July-Sep 14158 12556 24.57 1602 24.55
Oct-Dec 14222 12583 24.63 1639 25.12
Elixhauser comorbidity score 0 12900 12126 23.73 774 11.86
1 14092 12728 24.91 1364 20.9
2 11958 10555 20.66 1403 21.5
≥3 18671 15687 30.7 2984 45.73
Median hosehold income for patient’s ZIP code, based on current year Missing 1130 1015 1.99 115 1.76
0–25 percentile 11924 10585 20.72 1339 20.52
26–50 percentile 12835 11354 22.22 1481 22.7
51–75 percentile 14754 13050 25.54 1704 26.11
76–100 percentile 16978 15092 29.54 1886 28.9
Elective versus non-elective admission Missing 86 78 0.15 DS DS
Non elective 31134 27421 53.67 3713 56.9
Elective 26401 23597 46.18 2804 42.97
Patient State is the same as Hospital State No 5790 5232 10.24 558 8.55
Yes 51831 45864 89.76 5967 91.45
Volume ≤ 90th percentile (73 / year) 29969 26472 51.81 3497 53.59
Above 90th percentile 27652 24624 48.19 3028 46.41
Control/ownership of hospital Government, nonfederal 9557 8572 16.78 985 15.1
Private, not-profit 43724 38652 75.65 5072 77.73
Private, invest-own 4340 3872 7.58 468 7.17
Bed size of hospital Small 3393 3057 5.98 336 5.15
Medium 8710 7720 15.11 990 15.17
Large 45518 40319 78.91 5199 79.68
Teaching status Non teaching 11576 10183 19.93 1393 21.35
Teaching 46045 40913 80.07 5132 78.65
Metropolitan Other 18315 16163 31.63 2152 32.98
Major metropolitan 39306 34933 68.37 4373 67.02
HCUP used Emergency No 34818 31028 60.72 3790 58.08
Yes 22803 20068 39.28 2735 41.92
Disposition Missing 62 57 0.11 DS DS
Routine 35432 32007 62.64 3425 52.49
Other 22127 19032 37.25 3095 47.43
Open in a new tab
*

DS = Data suppressed in accordance with the HCUP/NRD guidelines

Association of depression with index admissions outcomes

Depression was associated with increased rates of neurological complications (OR 1.03–1.18, p = 0.003) and higher rates of non-routine disposition (OR 1.10–1.25, p < 0.0001). Of note, analysis of other risk factors revealed that hospitals with ≤ 90th percentile for procedural volume were associated with a higher rate of major complication (OR 1.35–1.68, p < 0.0001) upon index admission (OR 1.49–2.00, p < 0.0001). In addition, patients over the age of 59 years old (age 60–74 years: OR 1.23–1.69, p < 0.0001; 75+: OR 1.57–2.23, p < 0.0001) experienced higher rates of major complications. (Tables 3, 4, 5.)

Table 3:

Predictors of major complication upon index admission for malignant brain tumor craniotomy. Values for odds ratio, confidence interval, and p-value are included.

Variable OR 95% CI p-value
Depression Yes 0.956858 0.832851 1.099439 0.5342
No Ref
Other Psychiatric disease Yes 0.766822 0.662987 0.88692 0.0003
No Ref
Alcohol and drug abusedisorder Yes 0.706452 0.497828 1.002503 0.0517
No
Smoking Yes 0.801957 0.725931 0.885857 <.0001
No Ref
Age ≤ 44 Ref
45–59 1.104508 0.928857 1.313243 0.2607
60–74 1.441811 1.231459 1.687925 <.0001
>75 1.869928 1.568626 2.229105 <.0001
Sex Female Ref
Male 1.674643 1.517554 1.848177 <.0001
APRDRG Mortality risk Minor Ref
Moderate 2.709597 1.608175 4.565829 0.0002
Major 13.28187 7.950223 22.19129 <.0001
Extreme 69.95135 41.62912 117.5425 <.0001
LOS ≤ 3 days Ref
4–6 days 1.691643 1.266681 2.259176 0.0004
7–11 days 2.124847 1.615267 2.79519 <.0001
> 12 days 3.997623 3.055368 5.230464 <.0001
Elixhauser comorbidity score 0 Ref
1 1.677492 1.274304 2.208249 0.0002
2 2.069072 1.573496 2.721001 <.0001
≥3 3.912983 3.017112 5.074865 <.0001
Volume ≤ 90th percentile (73 / year) 1.508024 1.351885 1.682028 <.0001
Above 90th percentile Ref
Control/ownership of hospital Government, nonfederal Ref
Private, not-profit 0.985506 0.861828 1.126933 0.8313
Private, invest-own 1.417649 1.168826 1.719442 0.0004
Open in a new tab

Table 4:

Predictors of neurological complication upon index admission for malignant brain tumor craniotomy. Values for odds ratio, confidence interval, and p-value are included.

Variable OR 95% CI p-value
Depression Yes 1.104287 1.034067 1.179157 0.0031
No Ref
Smoking Yes 0.92118 0.878008 0.966378 0.0008
No Ref
Age ≤ 44 Ref
45–59 0.761321 0.715052 0.810584 <.0001
60–74 0.52482 0.488166 0.564226 <.0001
>75 0.394791 0.355653 0.438191 <.0001
Sex Female Ref
Male 1.264656 1.210702 1.321147 <.0001
LOS ≤ 3 days Ref
4–6 days 1.056858 0.986295 1.132469 0.1165
7–11 days 1.250196 1.155808 1.352426 <.0001
> 12 days 1.797685 1.652022 1.956388 <.0001
Elixhauser comorbidity score 0 Ref
1 0.970154 0.907193 1.037486 0.3764
2 0.990149 0.916036 1.070151 0.8024
≥3 1.102742 1.021222 1.19077 0.0125
Volume ≤ 90th percentile (73 / year) 0.80284 0.748338 0.861311 <.0001
Above 90th percentile Ref
Control/ownership of hospital Government, nonfederal Ref
Private, not-profit 1.2532 1.151425 1.364107 <.0001
Private, invest-own 1.256588 1.107937 1.42504 0.0004
Insurance Medicare 1.209854 1.111711 1.316662 <.0001
Medicaid Ref
Private insurance 1.10186 1.02296 1.186728 0.0105
Self-pay 1.056224 0.918512 1.214582 0.4425
No charge 0.955902 0.599955 1.52318 0.8496
Other 1.119072 0.984915 1.271503 0.0841
Discharge quarter Jan-March Ref
Apr-Jun 1.003707 0.944027 1.067159 0.9058
July-Sep 0.976969 0.918237 1.039459 0.4606
Oct-Dec 0.907284 0.850526 0.967732 0.0031
Median hosehold income for patient’s ZIP code, based on current year 0–25 percentile Ref
26–50 percentile 0.942895 0.881703 1.008234 0.0855
51–75 percentile 1.016434 0.946391 1.091661 0.6545
76–100 percentile 1.082962 1.008133 1.163229 0.029
Elective versus non-elective admission Non elective 0.656587 0.620456 0.694752 <.0001
Elective Ref
Teaching status Non teaching 1.102963 1.02891 1.182345 0.0057
Teaching Ref
Metropolitan Other 0.831936 0.780126 0.887275 <.0001
Major metropolitan Ref
Open in a new tab

Table 5:

Predictors of non-routine disposition upon discharge upon index admission for malignant brain tumor craniotomy. Values for odds ratio, confidence interval, and p-value are included.

Variable OR 95% CI p-value
Depression Yes 1.171166 1.100209 1.246575 <.0001
No Ref
Other psychiatric disease Yes 1.097242 1.027882 1.171283 0.0053
No Ref
Smoking Yes 0.890297 0.850271 0.932301 <.0001
No Ref
Age ≤ 44 Ref
45–59 1.87311 1.731867 2.025872 <.0001
60–74 2.641112 2.424439 2.877437 <.0001
>75 4.813382 4.31458 5.369313 <.0001
Sex Female Ref
Male 0.803884 0.772132 0.836942 <.0001
APRDRG Mortality risk Minor Ref
Moderate 1.203098 1.127046 1.284411 <.0001
Major 1.580751 1.467558 1.702674 <.0001
Extreme 2.533242 2.291943 2.799946 <.0001
LOS ≤ 3 days Ref
4–6 days 3.026177 2.815951 3.252097 <.0001
7–11 days 5.566687 5.120745 6.052068 <.0001
> 12 days 9.560117 8.70067 10.50446 <.0001
Elixhauser comorbidity score 0 Ref
1 1.271249 1.189342 1.358661 <.0001
2 1.389022 1.298098 1.486315 <.0001
≥3 1.650701 1.537719 1.771984 <.0001
Elective Elective Ref
Non elective 0.7719 0.730665 0.815544 <.0001
Control/ownership of hospital Government, nonfederal Ref
Private, not-profit 1.287239 1.162299 1.42561 <.0001
Private, invest-own 1.53526 1.347836 1.748748 <.0001
Insurance Medicare 1.897619 1.745428 2.06308 <.0001
Medicaid Ref
Private insurance 1.076161 0.994913 1.163927 0.0667
Self-pay 0.70321 0.601096 0.82267 <.0001
No charge 0.700192 0.448476 1.093299 0.117
Other 0.946769 0.830108 1.079826 0.4146
Hospital bedsize Small 0.712269 0.576085 0.880734 0.0017
Medium 0.970251 0.87293 1.078531 0.5762
Large Ref
Metropolitan Other 0.883027 0.817585 0.95361 0.0015
Major metropolitan Ref
Open in a new tab

30- and 90-day Readmissions demographics

There were 49,340 patients that met inclusion criteria for 30-day analysis, with 7,301 patients (%) readmitted within 30 days of index hospitalization. Of the 40,599 patients that met 90-day analysis inclusion criteria, 11,958 patients (%) were readmitted. 30- and 90- day readmissions shared the most frequent diagnoses upon readmission – postoperative infection (15.76% and 14.03%, respectively) or secondary malignant neoplasm of the brain or spine (7.4% and 8.2%, respectively) Table 6.

Table 6.

The most frequent primary diagnoses upon 30-day and 90-day readmission.

ICD-9 primary diagnosis Diagnosis Code Frequency (%)
30-day 90-day
Post-operative infection 998.59, 0389, 486, 5990 15.76 14.03
Secondary malignant neoplasm of brain/spine 198.3 7.4 8.18
Other pulmonary embolic infarct 41519 3.92 4.41
Malignant neoplasm of brain, unspecified 191.9 2.53 3.27
Malignant neoplasm of frontal lobe 1911 2.49 2.84
Other nervous system complications 997.09 1.99 1.28
Other convulsions 780.39 1.66 1.54
DVT/EMB of proximal lower extremity 453.41 1.6 1.61
Malignant neoplasm of other parts of brain 191.8 1.49 1.36
Malignant neoplasm of temporal lobe 191.2 1.48 1.94
Open in a new tab

Factors associated with 30-day readmission

Major (OR 1.06–1.27, p = 0.0012) and extreme (OR 1.18–1.49, p <0.0001) APRDRG disease severity risk ratings, as well as presence of 1 or more comorbidities, was associated with a greater likelihood of readmission. Patients with length of stay greater than 7 days and non-routine disposition upon discharge (OR 1.25–1.40, p < 0.0001) were more likely to be readmitted within 30-days. Patients undergoing non-elective craniotomies had decreased odds of readmission compared to those undergoing elective craniotomies (OR 0.80–0.94, p = 0.0003). There was no association between a comorbid diagnosis of depression and 30-day readmissions (OR 0.94–1.10, p < 0.6146). 30-day readmissions factors are further summarized in Table 7.

Table 7.

Predictors of 30-day all-cause readmissions for malignant brain tumor craniotomy. Values for odds ratio, confidence interval, and p-values are included.

Variable OR 95% CI p-value
Depression Yes 1.020303 0.943461 1.103404 0.6146
No Ref
Sex Female Ref
Male 1.187203 1.126708 1.251071 <.0001
APRDRG Mortality risk Minor Ref
Moderate 1.018978 0.940071 1.104508 0.6482
Major 1.160673 1.060775 1.270106 0.0012
Extreme 1.318903 1.177037 1.478015 <.0001
LOS ≤ 3 days Ref
4–6 days 1.081663 0.992925 1.17845 0.0723
7–11 days 1.213611 1.10949 1.327503 <.0001
> 12 days 1.392638 1.270614 1.526382 <.0001
Elixhauser comorbidity score 0 Ref
1 1.155808 1.06173 1.258222 0.0008
2 1.300046 1.191008 1.419068 <.0001
≥3 1.509532 1.38805 1.641811 <.0001
Elective Elective Ref
Non elective 0.863553 0.797 0.935756 0.0003
Resident of the state Yes Ref
No 0.675164 0.605077 0.753444 <.0001
Insurance Medicare 0.859074 0.791045 0.93286 0.0003
Medicaid Ref
Private insurance 0.768358 0.708291 0.833518 <.0001
Self-pay 0.711272 0.596426 0.848318 0.0002
No charge 0.946959 0.614774 1.458633 0.8046
Other 0.678684 0.575567 0.800275 <.0001
Metropolitan Other 0.866754 0.817177 0.919247 <.0001
Major metropolitan Ref
HCUP used Emergency Yes Ref
No 0.908192 0.841979 0.979513 0.0126
Disposition Routine Ref
Other 1.322072 1.248696 1.399759 <.0001
Open in a new tab

Factors associated with 90-day readmissions

Patients with a major APR DRG disease severity rating(OR 1.05–1.23, p = 0.0008), LOS greater than 7 days (7–11 days: 1.09–1.27, p < 0.0001; 12+: OR 1.14–1.35, p < 0.0001 ), Elixhauser score greater than 0 (1 comorbidity: OR 1.14–1.32, p < 0.0001; 2: OR 1.29–1.51, p < 0.0001; 3+: OR 1.52–1.77, p < 0.0001), or a non-routine disposition upon discharge (OR 1.16–1.28, p < 0.0001) were at increased risk for readmission at 90 days. Patients older than 45 years old had an 18–30% increased risk of readmission (age 45–59 years: OR 1.13–1.30, p < 0.0001; 60–74: OR 1.19–1.39, p < 0.0001; 75+: OR 1.07–1.30, p = 0.0009). 90-day readmission factors are further summarized in Table 8.

Table 8.

Predictors of 90-day all-cause readmissions for malignant brain tumor craniotomy. Values for odds ratio, confidence interval, and p-values are included.

Variable OR 95% CI p-value
Depression Yes 1.029836 0.961558 1.102963 0.4005
No Ref
Other psychiatric disease Yes 0.926631 0.869793 0.987084 0.0181
No Ref
Age ≤ 44 Ref
45–59 1.215433 1.133715 1.30304 <.0001
60–74 1.29072 1.194467 1.394589 <.0001
>75 1.179629 1.06844 1.302258 0.0011
Sex Female Ref
Male 1.083395 1.033447 1.135871 0.0009
APRDRG Mortality risk Minor Ref
Moderate 1.00441 0.940635 1.072508 0.895
Major 1.137804 1.054957 1.227034 0.0008
Extreme 1.093518 0.988764 1.20925 0.0818
LOS ≤ 3 days Ref
4–6 days 1.070365 1.002603 1.142821 0.0416
7–11 days 1.176801 1.094393 1.265415 <.0001
> 12 days 1.240978 1.144537 1.345681 <.0001
Elixhauser comorbidity score 0 Ref
1 1.224093 1.138487 1.316267 <.0001
2 1.397661 1.294339 1.509231 <.0001
≥3 1.636566 1.515583 1.767206 <.0001
Teaching status Teaching Ref
Non teaching 1.084805 1.024495 1.148779 0.0053
Resident of the state Yes Ref
No 0.543242 0.489388 0.602963 <.0001
Insurance Medicare 0.855987 0.787415 0.930531 0.0003
Medicaid Ref
Private insurance 0.720507 0.668112 0.776934 <.0001
Self-pay 0.628198 0.538537 0.732714 <.0001
No charge 0.752993 0.507428 1.117395 0.1588
Other 0.716555 0.623442 0.823493 <.0001
Metropolitan Other 0.903843 0.857272 0.952943 0.0002
Major metropolitan Ref
HCUP used Emergency Yes Ref
No 0.90529 0.859418 0.95361 0.0002
Disposition Routine Ref
Other 1.219572 1.158586 1.283769 <.0001
Open in a new tab

DISCUSSION

The current study reports nationwide epidemiological rates and outcomes associated with comorbid depression in patients undergoing craniotomy for malignant brain tumors and shows a clear association between MDD and postoperative complication rates and discharge status in this highly susceptible cohort. The prevalence of comorbid depression in cancer patients has been previously reported in the ambulatory setting. However, research describing prevalence in the inpatient peri-operative environment is scarce. The rate of MDD in this particular malignant brain tumor cohort was amongst the highest (11.3%) when compared to rates of depression in other major cancer types, including lung (11.5%), breast (10.3%), head and neck (9.3%), colorectal (8.1%), and prostate cancer (4.9%).10,11 Furthermore, these rates reflect medical record diagnoses recorded by administrative billing codes, which have a tendency to underestimate the true burden of MDD. In a systematic review of depression in glioma patients, the median frequency of physician-diagnosed MDD was 15%.6,12 However, patient-reported rates of depression have been as high as 93%, 6,12 demonstrating a notable disparity between patient perception of mood disturbances when compared to clinically diagnosed psychiatric disorders. Although concerning, this may represent an opportunity for improved screening and intervention in this high risk population.

The aforementioned review found that across the literature, depression was consistently associated with functional impairment, cognitive dysfunction, and reduced quality of life. 6 However, the pathogenesis of depression in brain tumors patients is not well understood. Prior literature suggests a multifactorial etiology of depression in these patients, including knowledge of prognosis, neurological deficits, chemotherapy, radiotherapy, cognitive decline, and exacerbation of underlying mental health disorders.13 Beyond the expected psychological distress of a cancer diagnosis,14 brain tumors are unique in that they may contribute to direct focal cerebral dysfunction and inherent neurodegenerative processes they may induce.15 In addition to the neurophysiological effects of brain tumor invasion on the surrounding native tissue, adjuvant chemotherapy and radiotherapy can also induce cerebral edema and necrosis in peri-tumoral tissue or accelerate vascular dementia.16 These neuropsychiatric effects can further contribute to the development of depression in malignant brain tumor patients.

In our study, MDD was not found to be directly associated with increased likelihood of readmission. However, depression was associated with adverse discharge disposition, such as to long-term assisted care facilities. When evaluating nonroutine discharge as an independent variable in multivariate analysis, however, it was noted to be associated with readmission. Discharge disposition is a readily utilized surrogate marker of functional status in administrative database studies.17 While the association between depression and worse functional status appears intuitive, a recent systematic review observed conflicting data regarding this association.6 In a study of 77 patients with primary brain tumors, Mainio A et al. found that preoperative depression diagnosis as defined by Beck Depression Index (BDI) was associated with lower functional status (Karnosfky Performance Status, KPS). Interestingly, this association was found to persist at 3 months and 1 year postoperatively, despite the cohort including lesions of both malignant and benign histology .18 By contrast, a retrospective cohort study of malignant astrocytomas by Gathinji et al. did not identify such an association.19 Notably, in the Gathinji et. al. study, the categorization of depression only captured patients that had been diagnosed by their primary care physician or a psychiatrist.

Depression was one of several factors associated with non-routine discharge disposition (our corollary for poor functional outcome) in patients surgically treated for malignant brain tumors. The other factors included older age, female sex, other psychiatric disease, other pre-existing comorbidities, hospital type (ownership, bed size, location), and prolonged length of stay which have previously been identified in the literature.20–22 Muhlestein et al. built a machine learning model to study predictors of discharge disposition in patients who underwent craniotomy for brain tumor in the Nationwide Inpatient Sample.20 In their analysis, predictors of non-routine disposition included older age, female sex, pre-existing comorbidities, and extended hospitalization. However, the association between depression and non-routine discharge persisted even after adjusting for these aforementioned factors in a robust multivariate analysis.

To the authors’ knowledge, this is the first study to identify an association between MDD and neurological complications during hospitalization for craniotomy for brain tumor. Tumor specifics, such as location, multi-focality and size > 4 cm have been associated with increased risk of depression.12,23 Specifically, anatomic involvement of the ventral frontal, temporoparietal, and limbic system have all been associated with worse mood states in patients with brain lesions.24 Thus, the higher rates of neurological complications in patients with MDD may reflect increased complexity of tumor surgery. Whether there is a causative relationship between depression and neurologic deficits remains speculative. However, from a molecular perspective, serotonin receptors have been shown to modulate glioma invasion and migration.25 Likewise, dysregulation in diurnal variation of cortisol in depression has been postulated to contribute to altered tumor catabolism.26

We did not evaluate the association between comorbid MDD and survival, as previous literature suggests these effects are most appreciable when examining long term survival, an outcome not represented in the NRD, which only follows patients across a single calendar year. It is notable that in a retrospective study of 1,015 patients undergoing resection of malignant astrocytoma, independent of age, WHO histological grade, extent of resection, and adjuvant temozolomide, preoperative diagnosis of depression was associated with decreased survival (RR – 1.41). 19 This difference, was most notable past 12 months and at 20 months, suggesting that the effects of depression are most prominent when evaluating long term survival. Similarly, in a 5-year study of 75 patients harboring solitary primary brain tumors, Mainio et. al. found depression to be a negative prognosticator of survival in patients with low grade gliomas. This did not hold true for those with a histologic diagnosis of high-grade glioma, reflecting the longer duration of survival in patients with low grade lesions (22.5 months versus 50.2 months).

The findings of our study within the context of prior literature suggest that diagnosis of depression is associated with surgical outcomes, including decreased functional status and postoperative neurological deficits, in patients with malignant brain tumors. Maximization of quality of life is paramount in these patients and is compromised by this potentially modifiable comorbidity. A multi-disciplinary and holistic approach to treating depression in brain tumor patients is essential., At present, several observational studies have found antidepressants to be under-prescribed in brain tumor patients (Armstrong). In the Glioma Outcomes Project, 93% of patients endorsed symptoms of depression, however only 15% of patients were diagnosed with clinical depression by their physicians, and of which only half (7%) were prescribed antidepressants.12 Recent preclinical studies found that fluoxetine, in addition to treating depression, was also selectively toxic to glioma cells.27 Many studies have also outlined non-pharmacological modalities for treating depression in brain tumor patients, including telephone-based support systems28, exercise programs29, and dyadic yoga30. In a pilot trial of 20 glioma patients and their caregivers, Milbury K et al., found that a 12-session dyadic yoga program resulted in clinically significant improvements in depression, cancer-related symptoms, and quality of life.30 Taken together, these data and ours support the integral role of mental health specialists, social workers and neuropsychologists into multidisciplinary brain tumor centers to provide routine screening and tailored treatment to this particularly vulnerable population.

Our study is subject to several limitations that are inherent to most large administrative database analyses. Particularly, the patient data identified in the NRD is limited to diagnoses and procedures identifiable through available ICD-9-CM codes. A large administrative dataset like the NRD does not have the granularity to capture all factors contributing to patient outcomes. Specifically, we could not identify several important clinical details including severity of depression (e.g., patient health questionnaire), pre-admission functional status, radiological imaging, peri-operative management, tumor specifics, surgical technique/approach variables or post-operative quality of life. This limits the number of confounding factors that we could adjust for in our analysis. Thus, the relationships between variables in our study are merely independent associations and cannot prove causality or temporality. The NRD patient linkage numbers do not track across calender years, which may slightly underestimate readmission rates. However, we addressed this limitation by excluding the last 1-month and 3-months from the index cohorts when evaluating 30-d and 90-d readmission, respectively. Some limitations were specific to this particular study design and patient population. The available ICD-9 codes do not distinguish between primary and secondary brain tumors, which limits the specificity of our results. Furthermore, the coding schemes cannot stratify by tumor grade, tumor size, or type of malignancy. Nevertheless, the high volume of cases and readmissions studied in this major nationwide database provide a high degree of generalizability, and selective insight into the relation between MDD and malignant brain tumor surgery and recovery.

CONCLUSION

Our findings support the hypothesis that depression is associated with poor surgical outcomes in patients undergoing craniotomy for resection of malignant brain tumors. Attention to this modifiable comorbidity during preoperative evaluation and patient recovery is crucial, and a routine screening and intervention strategy may improve patient-centered outcomes. Given the limitations of an administrative database study, further investigation is warranted in the form of a prospective study design. A future study may aim to identify indications for screening and treatment of depression in this population.

Acknowledgments

Funding: FJA is supported by a NIH SC CTSI KL2 Clinical and Translational Research Scholar Award.

Abbreviations:

AHRQ

Agency for Healthcare Research and Quality

APR-DRG

all patient refined diagnosis related groups

GBM

glioblastoma multiforme

HCUP

Healthcare Cost and Utilization Project

ICD-9CM

International Classification of Disease, Ninth Edition, Clinical Modification

LOS

length of stay

MDD

major depressive disorder

NRD

Nationwide Readmission Database

OR

Odds Ratio

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Disclosures: The authors of this manuscript have no financial or industry connections relevant to the contents of this manuscript.

Disclosure-Conflict of Interest: The authors of this manuscript have no financial or industry connections relevant to the contents of this manuscript. We report no potential conflicts of interest.

References

  • 1.Walker J, Hansen CH, Martin P, et al. Prevalence, associations, and adequacy of treatment of major depression in patients with cancer: a cross-sectional analysis of routinely collected clinical data. Lancet Psychiatry. 2014;1(5):343–350. doi: 10.1016/S2215-0366(14)70313-X [DOI] [PubMed] [Google Scholar]
  • 2.Honda K, Goodwin RD. Cancer and mental disorders in a national community sample: findings from the national comorbidity survey. Psychother Psychosom. 2004;73(4):235–242. doi: 10.1159/000077742 [DOI] [PubMed] [Google Scholar]
  • 3.Krebber AMH, Buffart LM, Kleijn G, et al. Prevalence of depression in cancer patients: a meta-analysis of diagnostic interviews and self-report instruments. Psychooncology. 2014;23(2):121–130. doi: 10.1002/pon.3409 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Omuro A, DeAngelis LM. Glioblastoma and other malignant gliomas: a clinical review. JAMA. 2013;310(17):1842–1850. doi: 10.1001/jama.2013.280319 [DOI] [PubMed] [Google Scholar]
  • 5.Tamimi AF, Juweid M. Epidemiology and Outcome of Glioblastoma In: De Vleeschouwer S, ed. Glioblastoma. Codon Publications; 2017. Accessed September 20, 2019 http://www.ncbi.nlm.nih.gov/books/NBK470003/ [PubMed] [Google Scholar]
  • 6.Rooney AG, Carson A, Grant R. Depression in cerebral glioma patients: a systematic review of observational studies. J Natl Cancer Inst. 2011;103(1):61–76. doi: 10.1093/jnci/djq458 [DOI] [PubMed] [Google Scholar]
  • 7.Anderson SI, Taylor R, Whittle IR. Mood disorders in patients after treatment for primary intracranial tumours. Br J Neurosurg. 1999;13(5):480–485. doi: 10.1080/02688699908540622 [DOI] [PubMed] [Google Scholar]
  • 8.Pelletier G, Verhoef MJ, Khatri N, Hagen N. Quality of Life in Brain Tumor Patients: The Relative Contributions of Depression, Fatigue, Emotional Distress, and Existential Issues. J Neurooncol. 2002;57(1):41–49. doi: 10.1023/A:1015728825642 [DOI] [PubMed] [Google Scholar]
  • 9.McCormick N, Bhole V, Lacaille D, Avina-Zubieta JA. Validity of Diagnostic Codes for Acute Stroke in Administrative Databases: A Systematic Review. PLoS ONE. 2015;10(8). doi: 10.1371/journal.pone.0135834 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Patel RS, Wen K-Y, Aggarwal R. Demographic Pattern and Hospitalization Outcomes of Depression among 2.1 Million Americans with Four Major Cancers in the United States. Med Sci Basel Switz. 2018;6(4). doi: 10.3390/medsci6040093 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Rohde RL, Adjei Boakye E, Challapalli SD, et al. Prevalence and sociodemographic factors associated with depression among hospitalized patients with head and neck cancer-Results from a national study. Psychooncology. 2018;27(12):2809–2814. doi: 10.1002/pon.4893 [DOI] [PubMed] [Google Scholar]
  • 12.Litofsky NS, Farace E, Anderson F, Meyers CA, Huang W, Laws ER. Depression in Patients with High-grade Glioma: Results of the Glioma Outcomes Project. Neurosurgery. 2004;54(2):358–367. doi: 10.1227/01.NEU.0000103450.94724.A2 [DOI] [PubMed] [Google Scholar]
  • 13.Pace A, Pompili A. Depression in Patients with High-grade Glioma: Results of the Glioma Project. Neurosurgery. 2005;56(3):E629–E629. doi: 10.1227/01.NEU.0000155090.77873.BE [DOI] [PubMed] [Google Scholar]
  • 14.Goebel S, Stark AM, Kaup L, von Harscher M, Mehdorn HM. Distress in patients with newly diagnosed brain tumours. Psychooncology. 2011;20(6):623–630. doi: 10.1002/pon.1958 [DOI] [PubMed] [Google Scholar]
  • 15.Weitzner MA. Psychosocial and neuropsychiatric aspects of patients with primary brain tumors. Cancer Invest. 1999;17(4):285–291; discussion 296–297. doi: 10.3109/07357909909040599 [DOI] [PubMed] [Google Scholar]
  • 16.Ali FS, Hussain MR, Gutiérrez C, et al. Cognitive disability in adult patients with brain tumors. Cancer Treat Rev. 2018;65:33–40. doi: 10.1016/j.ctrv.2018.02.007 [DOI] [PubMed] [Google Scholar]
  • 17.Qureshi AI, Chaudhry SA, Sapkota BL, Rodriguez GJ, Suri MFK. Discharge Destination as a Surrogate for Modified Rankin Scale Defined Outcomes at 3- and 12-Months Poststroke Among Stroke Survivors. Arch Phys Med Rehabil. 2012;93(8):1408–1413.e1. doi: 10.1016/j.apmr.2012.02.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Mainio A, Hakko H, Timonen M, Niemelä A, Koivukangas J, Räsänen P. Depression in relation to survival among neurosurgical patients with a primary brain tumor: a 5-year follow-up study. Neurosurgery. 2005;56(6):1234–1241; discussion 1241–1242. doi: 10.1227/01.neu.0000159648.44507.7f [DOI] [PubMed] [Google Scholar]
  • 19.Gathinji M, McGirt MJ, Attenello FJ, et al. Association of preoperative depression and survival after resection of malignant brain astrocytoma. Surg Neurol. 2009;71(3):299–303, discussion 303. doi: 10.1016/j.surneu.2008.07.016 [DOI] [PubMed] [Google Scholar]
  • 20.Muhlestein WE, Akagi DS, Chotai S, Chambless LB. The impact of presurgical comorbidities on discharge disposition and length of hospitalization following craniotomy for brain tumor. Surg Neurol Int. 2017;8. doi: 10.4103/sni.sni_54_17 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Barker FG, Curry WT, Carter BS. Surgery for primary supratentorial brain tumors in the United States, 1988 to 2000: The effect of provider caseload and centralization of care. Neuro-Oncol. 2005;7(1):49–63. doi: 10.1215/S1152851704000146 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.De la Garza-Ramos R, Kerezoudis P, Tamargo RJ, Brem H, Huang J, Bydon M. Surgical complications following malignant brain tumor surgery: An analysis of 2002–2011 data. Clin Neurol Neurosurg. 2016;140:6–10. doi: 10.1016/j.clineuro.2015.11.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Armstrong TS, Cohen MZ, Eriksen LR, Hickey JV. Symptom clusters in oncology patients and implications for symptom research in people with primary brain tumors. J Nurs Scholarsh Off Publ Sigma Theta Tau Int Honor Soc Nurs. 2004;36(3):197–206. doi: 10.1111/j.1547-5069.2004.04038.x [DOI] [PubMed] [Google Scholar]
  • 24.Irle E, Peper M, Wowra B, Kunze S. Mood changes after surgery for tumors of the cerebral cortex. Arch Neurol. 1994;51(2):164–174. doi: 10.1001/archneur.1994.00540140070017 [DOI] [PubMed] [Google Scholar]
  • 25.Merzak A, Koochekpour S, Fillion MP, Fillion G, Pilkington GJ. Expression of serotonin receptors in human fetal astrocytes and glioma cell lines: a possible role in glioma cell proliferation and migration. Brain Res Mol Brain Res. 1996;41(1–2):1–7. doi: 10.1016/0169-328x(96)00058-7 [DOI] [PubMed] [Google Scholar]
  • 26.Spiegel D, Giese-Davis J. Depression and cancer: mechanisms and disease progression. Biol Psychiatry. 2003;54(3):269–282. doi: 10.1016/s0006-3223(03)00566-3 [DOI] [PubMed] [Google Scholar]
  • 27.Liu K-H, Yang S-T, Lin Y-K, et al. Fluoxetine, an antidepressant, suppresses glioblastoma by evoking AMPAR-mediated calcium-dependent apoptosis. Oncotarget. 2014;6(7):5088–5101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Jones S, Ownsworth T, Shum DHK. Feasibility and Utility of Telephone-Based Psychological Support for People with Brain Tumor: A Single-Case Experimental Study. Front Oncol. 2015;5. doi: 10.3389/fonc.2015.00071 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Cordier D, Gerber M, Brand S. Effects of two types of exercise training on psychological well-being, sleep, quality of life and physical fitness in patients with high-grade glioma (WHO III and IV): study protocol for a randomized controlled trial. Cancer Commun Lond Engl. 2019;39(1):46. doi: 10.1186/s40880-019-0390-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Milbury K, Li J, Weathers S-P, et al. Pilot randomized, controlled trial of a dyadic yoga program for glioma patients undergoing radiotherapy and their family caregivers. Neuro-Oncol Pract. 2019;6(4):311–320. doi: 10.1093/nop/npy052 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES