Abstract
BACKGROUND
Research on age-related complications secondary to shunts in normal pressure hydrocephalus (NPH) is primarily limited to single-center studies and small cohorts.
OBJECTIVE
To determine the rates of hospital readmission and surgical complications, and factors that predict them, following shunt surgery for NPH in a large healthcare network.
METHODS
Surgical procedures, complications, and readmissions for adults undergoing ventricular shunting for NPH were determined using de-identified claims from a privately insured United States healthcare network in years 2007-2014. Univariate and multivariate statistics were used to determine factors that predict poor surgical outcomes. The primary outcome variable was surgical complications or readmissions (composite variable for any major perioperative complication or 30-d readmission).
RESULTS
The 30-d readmission rate for 974 patients with NPH who underwent ventricular shunting was 7.29%; the most common reasons for readmission were shunt-related complications, infection, hemorrhage, altered mental status, and cardiopulmonary and musculoskeletal problems. The perioperative complication rate was 21.15%, including intraparenchymal hemorrhage (5.85%) and extra-axial (subdural or epidural) hematoma (5.54%). The overall rate of having a surgical complication or 30-d readmission was 25.15%. Age did not predict surgical complication or 30-d readmission. Preoperative comorbidities independently associated with poor outcome were myocardial infarction within 1 yr (OR = 3.984, 95% CI = 1.105-14.368); existing cerebrovascular disease (odds ratio [OR] = 2.206, 95% CI = 1.544-3.152); and moderate/severe renal disease (OR = 2.000, 95% CI = 1.155-3.464).
CONCLUSION
The rate of complications or readmission within 30 d of ventricular shunting for NPH is 25.15%. Preoperative comorbidities of myocardial infarction within 1 yr, cerebrovascular disease, and moderate/severe renal disease are independent risk factors for poor outcome.
Keywords: Complications, Normal pressure hydrocephalus, Prediction, Shunt, Surgical outcomes
Graphical Abstract
Graphical Abstract.
ABBREVIATIONS
- CI
confidence interval
- CPT
Current Procedural Terminology
- CVD
cerebrovascular disease
- ICD
International Classification of Diseases, 9th revision
- MI
myocardial infarction
- NPH
normal pressure hydrocephalus
- OR
odds ratio
- PVD
peripheral vascular disease
Normal pressure hydrocephalus (NPH) was originally described in 1957 by Dr Salómon Hakim2 as cerebral ventriculomegaly and a triad of clinical symptoms, including gait abnormalities, dementia, and urinary incontinence.1-3 Today, NPH is recognized as a disease that primarily affects older adults.4-8 The placement of ventriculoperitoneal or ventriculoatrial shunts is a well-established treatment to alleviate the symptoms of NPH.8-12
Multiple studies, however, have demonstrated that complications from the placement of shunts are relatively common, with some reporting complication rates as high as 42.5%.8-10,13,14 As has been demonstrated in other surgical fields, older patients are particularly susceptible to postoperative compications.15-17
Because NPH is primarily a disorder of older adults, investigators have explored associations between older age and the high complication rates seen in patients with NPH treated with shunt placement.8 The available literature on age-related complications secondary to shunts in NPH, however, is primarily limited to single-center studies with relatively small cohorts.
In this study, we used a United States medical insurance claims-based sample of NPH patients with newly placed ventricular shunts to examine patient-level factors that may predict having a surgical complication or hospital readmission within 30 d of ventricular shunting for NPH.
METHODS
The Clinformatics™ Data Mart database (Optum®, Eden Prairie, Minnesota) contains de-identified claims from a large, private healthcare network of over 58 million beneficiaries in the United States from 2001 to 2014. Using both International Classification of Diseases, 9th revision (ICD-9) diagnosis codes and Current Procedural Terminology (CPT) codes, all enrollees with both an inpatient and outpatient diagnosis claim for NPH (331.5) were identified. For each identified NPH case, we documented associated medical and surgical procedures throughout their period of longest insurance coverage.
For each patient, start and end dates for all enrollment periods were available. In this analysis, we examined the longest period of continuous coverage that was not interrupted by more than a 30-d gap. In addition to the de-identified medical and surgical claims data, basic demographic information was available for all patients. Because the data were de-identified, the University of Michigan Institutional Review Board considered this to be a nonregulated study, and informed consent was not required.
Identification of Patients With NPH
The patient selection algorithm for this study is depicted in the Figure. The Clinformatics™ Data Mart (Optum®, Eden Prairie, Minnesota) was searched for all enrollees with an ICD-9-CM diagnosis code of 331.5 (NPH) from January 1, 2007 to December 31, 2014. Of note, the ICD-9-CM diagnosis code for NPH (331.5) was not put into use until 2007. All outpatient medical claims and inpatient admissions were pulled for the list of unique patients with an NPH code. Enrollees with an outpatient medical claim but no inpatient admission data were excluded. The data were then queried for enrollees who also had a CPT code for ventricular shunting (62220 or 62223). Those without both an NPH diagnosis code and ventricular shunt code were excluded. Additional exclusion criteria were employed to be sure that the enrollees selected had sufficient data for analysis, and that the NPH diagnosis code and CPT shunt code were for the index diagnosis and shunt surgery. This meant excluding enrollees whose first coding for NPH was not before or within 90 d of the ventricular shunt coding, whose ventricular shunt coding was not during their period of longest coverage, and who did not have at least 1 yr of continuous coverage preceding the first NPH code. These selection criteria were chosen to increase the overall accuracy of our claims-based NPH case ascertainment.
FIGURE.
Identification algorithm for patients with normal pressure hydrocephalus.
Identifying Surgical Complications
Surgical complications were identified using a series of ICD-9 codes (Table, Supplemental Digital Content 1). Only complications that were identified in the first 30 d after ventricular shunt surgery were included in the analysis. The 30-d complications included ventricular shunt infection, cerebrospinal fluid leak, ischemic stroke, intracerebral hemorrhage, epidural or subdural hematoma, any intracranial hemorrhage (composite of intracerebral hemorrhage and epidural or subdural hematoma), seizure, coma, myocardial infarction (MI), deep vein thrombosis, pulmonary embolism, pneumothorax, infectious pneumonia, aspiration pneumonia, meningitis or encephalitis, sepsis or septic shock, ileus, wound infection, visceral injury, peritonitis, and other central nervous system complications.
Outcome Measures and Predictors of Surgical Complications and Hospital Readmissions
The primary outcome variable in this study was 30-d surgical complications or hospital readmissions, which was a composite variable defined as the presence of a code for at least one of the surgical complications outlined above, or a readmission to the hospital within 30 d of discharge for the index shunt surgery.18 A person was considered to be readmitted to the hospital if he or she had a hospital admission code between 1 and 30 d after a hospital discharge code. Admissions on the same day as discharge from the index hospitalization were considered transfers or admission to rehabilitation facilities and were excluded. Discharges to home, with or without nursing care, were identified via discharge destination codes.
A number of antecedent factors were evaluated as potential predictors for surgical complications or readmissions in the year preceding surgery. Age at the time of surgery was evaluated as a categorical variable, stratified by decade of life (<60 yr old, 60-70 yr old, 70-80 yr old, and 80+ yr old). Additional factors were individual comorbidities of the Charlson Comorbidity Index, a validated tool for identifying and assessing the relative severity of patient comorbidities from administrative data.19,20 Individual Charlson comorbidities identified included the following: MI within the 1 yr preceding surgery, history of MI, congestive heart failure, peripheral vascular disease (PVD), cerebrovascular disease (CVD), chronic obstructive pulmonary disease, dementia, paralysis, diabetes, diabetes with complications, renal disease, mild liver disease, moderate/severe liver disease, peptic ulcer disease, rheumatologic disease, and acquired immunodeficiency syndrome. Hypertension, although not a Charlson comorbidity, is highly prevalent in this population and was therefore also included in this analysis.
Undergoing a cranial reoperation within 30 d was used as a surrogate for understanding the severity of intracranial complications after ventricular shunt surgery. As such, the database was queried to determine how many enrollees underwent one of the following surgical interventions within 30 d of the index shunt surgery: revision or replacement of ventricular shunt catheter; craniotomy for evacuation of epidural or subdural hematoma; craniotomy for evacuation of intracerebral hemorrhage; burr hole craniotomy for hematoma evacuation; and exploratory craniotomy or craniectomy. The same antecedent factors stated above were evaluated as potential predictors of the need for a cranial reoperation within 30 d.
Statistical Analysis
All data were analyzed using the SAS 9.4 statistical software (SAS Institute Inc., Cary, North Carolina). Results were exported and compiled in Excel software (Microsoft, Redmond, Washington). Descriptive statistics were assessed for significance using chi-square tests as appropriate at the alpha = 0.05 level. Univariate and multivariate logistic regressions were used for predictive modeling of the impact of each factor on the composite outcome variable. In the multivariate logistic regression model, a stepwise approach was used. The parameter for variable entry into the model was an alpha ≤0.20, and an alpha ≤0.05 was required for variable retention in the model. Bivariate and stepwise logistic regressions, as described above, were conducted to analyze factors that may predict a surgical complication or readmission, or the need for a cranial reoperation within 30 d.
RESULTS
Demographic Information
A total of 974 enrollees met the criteria for inclusion into this study (Figure). Basic characteristics of the cohort are displayed in Table 1. The average length of coverage among those included in the study was 6.61 ± 3.02 yr.
TABLE 1.
Characteristics of Enrollees With Shunted Normal Pressure Hydrocephalus
| Count or mean | % or SD | Median (IQR) | |
|---|---|---|---|
| Total number (N) | 974 | ||
| Average length of coverage (years) | 6.61 | 3.02 | 6.01 (4.09-8.92) |
| Male gender | 548 | 56.26% | |
| Race | |||
| Asian | 13 | 1.33% | |
| Black | 49 | 5.03% | |
| Hispanic | 45 | 4.62% | |
| Unknown | 50 | 5.13% | |
| White | 776 | 79.67% | |
| Missing | 41 | 4.21% | |
| Age | |||
| <60 yr | 77 | 7.91% | |
| 60-70 yr | 164 | 16.84% | |
| 70-80 yr | 411 | 42.20% | |
| 80+ yr | 322 | 33.06% | |
| Charlson Comorbidity Index score | 1.03 | 1.23 | 1.00 (0.00-2.00) |
| MI in year preceding surgery | 11 | 1.13% | |
| History of MI | 14 | 1.44% | |
| Congestive heart failure | 56 | 5.75% | |
| Peripheral vascular disease | 21 | 2.16% | |
| Cerebrovascular disease | 166 | 17.04% | |
| Chronic obstructive pulmonary disease | 89 | 9.14% | |
| Dementia | 179 | 18.38% | |
| Paralysis | 19 | 1.95% | |
| Diabetes mellitus (DM) | 207 | 21.25% | |
| DM complicated | 22 | 2.26% | |
| Moderate/severe renal disease | 61 | 6.26% | |
| Mild liver disease | 5 | 0.51% | |
| Moderate/severe liver disease | 2 | 0.21% | |
| Peptic ulcers | 8 | 0.82% | |
| Rheumatologic disease | 19 | 1.95% | |
| Acquired immunodeficiency syndrome | 2 | 0.21% | |
| Hypertension | 438 | 44.97% | |
IQR, interquartile range; MI, myocardial infarction; SD, standard deviation.
Postoperative Complications and Readmissions
Basic postoperative outcomes are displayed in Table 2. Following the index ventricular shunt placement, the median (interquartile range) length of stay for the overall cohort was 3.00 (6.00) d. The overall 30-d readmission rate was 7.29% (n = 71). The most common reasons for readmission were shunt-related complications, infection, hemorrhage, delirium or altered mental status, cardiopulmonary complications, and musculoskeletal problems (Table, Supplemental Digital Content 2). There was no significant difference in readmission rate based on age at the time of surgery (P = .44). The overall rate of at least one perioperative complication was 21.15% (n = 206). There was no significant difference in perioperative complication rates based on age at the time of surgery (P = .09).
TABLE 2.
Perioperative Complications
| Age stratified | ||||||
|---|---|---|---|---|---|---|
| Overall cohort | <60 yr | 60-70 yr | 70-80 yr | 80+ yr | P value | |
| Primary outcome: surgical complication or readmission | 245 (25.15%) | 21 (27.27%) | 33 (20.12%) | 92 (22.38%) | 99 (30.75%) | .024 |
| Length of stay (days), median (IQR) | 3.00 (6.00) | 3.00 (6.00) | 3.00 (4.00) | 3.00 (5.00) | 4.00 (5.00) | |
| Thirty-day readmission rate | 71 (7.29%) | 9 (11.69%) | 8 (4.88%) | 27 (6.57%) | 27 (8.39%) | .44 |
| Discharged to home (n = 860) | 559 (65.0%) | 52 (81.25%) | 110 (75.86%) | 235 (62.83%) | 162 (58.48%) | <.0001 |
| Perioperative complications (N (%)) | ||||||
| Any perioperative complication | 206 (21.15%) | 15 (19.48%) | 32 (19.51%) | 76 (18.49%) | 86 (25.78%) | .09 |
| Shunt infection | 23 (2.36%) | 4 (5.19%) | 5 (3.05%) | 8 (1.95%) | 6 (1.86%) | .30 |
| Cerebrospinal fluid leak | 1 (0.10%) | 0 (0.00%) | 0 (0.00%) | 1 (0.24%) | 0 (0.00%) | .71 |
| Stroke | 26 (2.67%) | 4 (5.19%) | 6 (3.66%) | 6 (1.46%) | 10 (3.11%) | .17 |
| Intracerebral hemorrhage | 57 (5.85%) | 5 (6.49%) | 10 (6.10%) | 21 (5.11%) | 21 (6.52%) | .86 |
| Epidural/subdural hematoma | 54 (5.54%) | 2 (2.60%) | 10 (6.10% | 21 (5.11%) | 21 (6.52%) | .55 |
| Any intracranial bleed | 87 (8.93%) | 5 (6.49%) | 15 (9.15%) | 34 (8.27%) | 33 (10.25%) | .69 |
| Seizure | 34 (3.49%) | 2 (2.60%) | 6 (3.66%) | 15 (3.65%) | 11 (3.42%) | .97 |
| Coma | 2 (0.21%) | 0 (0.00%) | 0 (0.00%) | 2 (0.49%) | 0 (0.00%) | .43 |
| Deep vein thrombosis | 16 (1.64%) | 0 (0.00%) | 3 (1.83%) | 7 (1.70%) | 6 (1.86%) | .70 |
| Pulmonary embolism | 5 (0.51%) | 1 (1.30%) | 0 (0.00%) | 0 (0.00%) | 4 (1.24%) | .06 |
| Pneumothorax | 1 (0.10%) | 0 (0.00%) | 1 (0.61%) | 0 (0.00%) | 0 (0.00%) | .18 |
| Infectious pneumonia | 8 (0.82%) | 1 (1.30%) | 2 (1.22%) | 1 (0.24%) | 4 (1.24%) | .40 |
| Aspiration pneumonia | 17 (1.75%) | 2 (2.60%) | 4 (2.44%) | 4 (0.97%) | 7 (2.17%) | .47 |
| Meningitis/encephalitis | 16 (1.64%) | 4 (5.19%) | 2 (1.22%) | 8 (1.95%) | 2 (0.62%) | .04 |
| Sepsis or septic shock | 18 (1.85%) | 3 (3.90%) | 5 (3.05%) | 4 (0.97%) | 6 (1.86%) | .19 |
| Postoperative myocardial infarction | 9 (0.92%) | 1 (1.30%) | 0 (0.00%) | 2 (0.49%) | 6 (1.86%) | .13 |
| Ileus | 16 (1.64%) | 2 (2.60%) | 2 (1.22%) | 6 (1.46%) | 6 (1.86%) | .85 |
| Wound infection | 8 (0.82%) | 1 (1.30%) | 1 (0.61%) | 3 (0.73%) | 3 (0.93%) | .94 |
| Visceral injury | 8 (0.82%) | 2 (2.60%) | 0 (0.00%) | 2 (0.29%) | 4 (1.24%) | .13 |
| Peritonitis | 2 (0.21%) | 0 (0.00%) | 0 (0.00%) | 1 (0.24%) | 1 (0.31%) | .87 |
| Other CNS complication | 5 (0.51%) | 0 (0.00%) | 0 (0.00%) | 3 (0.73%) | 2 (0.62%) | .64 |
CNS, central nervous system; IQR, interquartile range.
The primary outcome variable for this study was surgical complications or readmissions. This was a composite variable consisting of the presence of at least one perioperative surgical complication or a 30-d readmission. The overall rate of having a surgical complication or readmission within 30 d of the index surgery was 25.15% (n = 245). There was a significant difference among age groups in the rate complications and readmissions (P = .024). However, this relationship was nonmonotonic, with those older than 80 yr and those younger than 60 yr having the highest rates (30.75% and 27.27%, respectively).
Overall, 65% of enrollees were discharged home following their inpatient stay for ventricular shunting (n = 559). When stratified by age, there was a significant decrease in the rate of discharge to home with each increasing decade of age (P < .0001).
Perioperative Complication Rates and Rates of Cranial Reoperation Within 30 d
Perioperative complication rates were calculated for the overall cohort and were also stratified by age (Table 2). In the overall cohort, the most common perioperative complications were intracerebral hemorrhage (5.85%) and extra-axial (subdural or epidural) hematoma formation (5.54%). Any intracranial hemorrhage was a composite variable consisting of anyone with either an intracerebral hemorrhage or a subdural or epidural hematoma formation. In the overall cohort, 8.93% (n = 87) of enrollees experienced some type of intracranial hemorrhage. This indicates that some patients experienced both an intracerebral hemorrhage and an extra-axial hematoma.
Given the rate of intracranial complication described above, rates of reoperation within 30 d were calculated as a surrogate for understanding the severity of the intracranial complication. Cranial reoperations within 30 d of the index ventricular shunt surgery are described in Table 3. Overall, 57 patients (5.9%) underwent 60 total reoperations within 30 d of the index shunt surgery. The majority of these reoperations (50 of 60, 83.3%) were for revision or replacement of the ventricular shunt catheter. As stated above, 87 patients were coded as having an intracranial hemorrhagic complication, but only 10 (11.5%) underwent reoperation within 30 d for hematoma evacuation. The patient-level factors that may be associated with risk for needing reoperation are described in Table, Supplemental Digital Content 3.
TABLE 3.
Reoperations Within 30 d of Index Shunt Surgery
| Procedure | CPT code | Number of cases |
|---|---|---|
| Revision or replacement of shunt | 62225, 62230, 62256, 62258 | 50 (5.13%) |
| Craniotomy for evacuation of epidural or subdural hematoma | 61312 | 7 (0.72%) |
| Burr hole craniotomy for evacuation of hematoma | 61154 | 2 (0.21%) |
| Craniotomy for evacuation of intracerebral hemorrhage | 61313 | 1 (0.10%) |
Modeling Factors That Predict 30-d Surgical Complications and Readmissions
Having a 30-d surgical complication or readmission was a binary composite outcome variable. Any enrollee with a code for one of the perioperative complications above or a 30-d readmission was coded as having the primary outcome. Unadjusted logistic regression analyses were conducted examining the bivariate relationships between each hypothesized comorbidity and surgical complication or readmission (Table 4). The following preexisting comorbidities were significantly associated with having a surgical complication or readmission in their bivariate models (odds ratio [OR], 95% CI): MI within the year preceding shunt surgery (5.330, 1.547-18.367); PVD (3.380, 1.417-8.059); CVD (2.319, 1.630-3.299); paralysis (2.741, 1.101-6.827); diabetes (1.449, 1.033-2.033); and moderate/severe renal disease (2.354, 1.386-3.998).
TABLE 4.
Bivariate Logistic Regression Models of Factors That Predict Surgical Complication or Readmission
| Comorbidity | Odds ratio | 95% CI |
|---|---|---|
| Age at time of surgery | ||
| Less than 60 yr | 1.00 [Reference] | |
| 60-70 yr | 0.672 | 0.358-1.262 |
| 70-80 yr | 0.769 | 0.443-1.336 |
| 80+ yr | 1.184 | 0.680-2.061 |
| Myocardial infarction (MI) in year preceding surgery | 5.330a | 1.547-18.367 |
| History of MI | 1.193 | 0.371-3.840 |
| Congestive heart failure | 1.442 | 0.807-2.577 |
| Peripheral vascular disease | 3.380a | 1.417-8.059 |
| Cerebrovascular disease | 2.319a | 1.630-3.299 |
| Chronic obstructive pulmonary disease | 1.181 | 0.726-1.921 |
| Dementia | 1.321 | 0.922-1.893 |
| Paralysis | 2.741a | 1.101-6.827 |
| Diabetes | 1.449 a | 1.033-2.033 |
| Diabetes with complications | 1.737 | 0.842-3.585 |
| Moderate/severe renal disease | 2.354a | 1.386-3.998 |
| Mild liver disease | 0.743 | 0.083-6.678 |
| Moderate/severe liver disease | <0.0001 | <0.0001→999.999 |
| Peptic ulcer disease | 0.992 | 0.199-4.946 |
| Rheumatologic disease | 0.79 | 0.260-2.403 |
| Acquired immunodeficiency syndrome | <0.0001 | <0.0001→999.999 |
| Hypertension | 1.325 | 0.991-1.771 |
aSignificance at .05 level.
After examining the bivariate relationships, a stepwise multivariate logistic regression model was performed. The final predictive model is displayed in Table 5. The following variables were determined to be the significant predictors of having a surgical complication or readmission within 30 d of ventricular shunting for NPH (OR, 95% CI): MI within the year preceding shunt surgery (3.984, 1.105-14.368); CVD (2.206, 1.544-3.152); and moderate/severe renal disease (2.000, 1.155-3.464).
TABLE 5.
Stepwise Multivariate Regression Model of Predictors of Surgical Complication or Readmission
| Variable | Odds ratio | 95% CI |
|---|---|---|
| Myocardial infarction in year preceding surgery | 3.984a | 1.105-14.368 |
| Cerebrovascular disease | 2.206a | 1.544-3.152 |
| Moderate/severe renal disease | 2.000a | 1.155-3.464 |
aSignificance at .05 level.
Modeling Factors That Predict the Need for Cranial Reoperation Within 30 d
The same methodology as described in the previous section was used to assess factors that may predict the need for cranial reoperation within 30 d of the index ventricular shunting procedure. The binary composite outcome variable in this case, though, was whether an enrollee had a cranial reoperation within 30 d of the index shunt procedure. These cranial reoperations were either craniotomy for evacuation of epidural or subdural hematoma, craniotomy for evacuation of intracerebral hemorrhage, burr hole craniotomy for evacuation of a hematoma, exploratory craniotomy or craniectomy, or revision or replacement of ventricular shunt catheter. The bivariate relationships are described in Table, Supplemental Digital Content 4. The only preexisting comorbidity or personal characteristic associated, in a bivariate fashion, with having a cranial reoperation within 30 d was the presence of peptic ulcer disease (OR = 5.521, 95% CI = 1.089-27.991).
A stepwise multivariate logistic regression model was performed in the same manner as described in the previous section. The only variable that independently predicted having a cranial reoperation within 30 d of the index shunt surgery was having peptic ulcer disease.
DISCUSSION
We present data from a large United States cohort of NPH patients confirming a high rate of postoperative surgical complications and readmissions following ventricular shunt surgery. The primary outcome variable was surgical complications or readmissions, defined as having a 30-d readmission or perioperative complication code. Surgical complications or readmissions were noted in 25% of patients, and the overall need for reoperation within 30 d was approximately 5.9%. This was driven largely by reoperations for revision or replacement of the ventricular shunt catheter. Nearly 9% of patients in this cohort had a code for some type of intracranial hemorrhagic complication, although only 11.5% of those patients underwent reoperation for hematoma evacuation within 30 d. MI within the year preceding ventricular shunting, CVD, and moderate/severe renal disease were associated with having a major perioperative complication or 30-d readmission. Age was not independently associated with the primary outcome.
It has been previously reported that ventricular shunt procedures are associated with a high complication rate,12,21,22 though some studies suggest that shunting for NPH is associated with fewer complications than shunting for other reasons.12 For surgical decision making and risk stratification in an elderly population, it remains important to understand the factors that may predict postsurgical complications and readmissions in patients with NPH.
In our privately insured population, 25% of patients undergoing ventricular shunting for NPH experienced at least one complication or a 30-d readmission. Further, 8.9% of NPH patients undergoing an elective shunt procedure had a code for an intracranial hemorrhage. Though the nature of the dataset precluded analysis of the severity of the intracranial hemorrhage or differentiation between symptomatic and asymptomatic hemorrhage, we used the need for reoperation within 30 d as a surrogate for the severity of an intracranial complication. Why did only 11.5% of patients with a code for an intracranial hemorrhage undergo reoperation for evacuation of hematoma? It is possible that individuals who developed a subdural hygroma because of early overdrainage of their shunt catheter were coded as having a subdural hematoma, because there is not an ICD-9 code for subdural hygroma. It is also possible that individuals with clinically insignificant, or less significant, intracranial bleeding were still coded as having an intracranial hemorrhage, despite the fact that they did not need decompressive intervention. We hypothesize that it is likely a combination of those 2 factors.
We also conducted multivariate logistic regression modeling to analyze factors that may predict the need for cranial reoperation within 30 d of the index shunt surgery. The only significant independent predictor of the need for a cranial reoperation was having comorbid peptic ulcer disease. Given that there is no theoretical, physiological, or clinical reason that peptic ulcer disease might predict the need for cranial reoperation after ventricular shunting, we hypothesize that this finding is due to chance.
Our overall 30-d complication and reoperation rates are lower than or similar to those of other single-center series for shunt placement. Schenker et al12 reported that 58% of patients with NPH in their series experienced some types of surgical complication, although only approximately half of those complications required reoperation. Rinaldo et al22 reported a major complication rate of 17.1% following shunting for high-grade glioma, and in the patient population of Korinek et al,21 23.3% of patients required a reoperation for shunting for any reason. Examining intracranial hemorrhage specifically, IPH rates ranged from 4 to 10% and varied based on the indication for shunting.9,23,24 Another study of shunting in NPH found a subdural hematoma incidence of 7.7%.9 Our study did not demonstrate a significant increase in risk of a hemorrhagic complication with aging.
Multivariate analysis was employed to understand the relative importance of preoperative comorbidities in predicting complications and readmissions after ventricular shunting for NPH. We demonstrate that MI in the 1 yr preceding ventricular shunting, CVD, and moderate/severe renal disease are significant risk factors for having a postoperative complication or readmission. Age at the time of shunt surgery was not a significant predictor of having a surgical complication or readmission within 30 d. Interestingly, the relationship between age and surgical complications was nonlinear. The 2 age groups with the highest complication rates were those under 60 yr old and those over 80 yr old. A possible explanation is that patients under age 60 who are coded as having NPH may have a different disease process than an older patient coded with NPH. Although a diagnosis of NPH in patients younger than 60 yr of age would be distinctly unusual, this age group was maintained in an effort to discern whether there was an association between age and complications or medical condition and complications. These data clearly demonstrate that medical factors, and not age, are associated with complications. Despite a lack of association between decade of age and having a complication or readmission, enrollees who had ventricular shunt surgery at older ages were significantly less likely to be discharged to home than their younger counterparts. The association between age and discharge destination after hospitalizations for other conditions has been well described.25-28
Limitations
There are a number of limitations to studies that use insurance claims databases, as described elsewhere.29,30 Identification of NPH and shunting relies on ICD and CPT codes, and the presence of coding errors may bias the results. Though the NPH-specific ICD-9-CM code (331.5) has not been validated in administrative data previously, its specificity and the coupling of an NPH code with a CPT shunt code lends validity to its use, especially in those who were 60 yr and older included in the study.13 In using only this specific code, however, we may have missed a portion of NPH patients who were coded with a more generic hydrocephalus code. Fortunately, the sample size of our cohort limits the relative influence of any single coding error on the results.
CONCLUSION
Using a large, national insurance claims database, the population-level rate of having a 30-d readmission or perioperative complication following NPH shunt surgery was 25%. There was a nearly 9% rate of having a code for some type of intracranial hemorrhagic complication in this cohort, although only 11.5% of those individuals underwent hematoma evacuation within 30 d. MI in the 1 yr preceding ventricular shunt surgery, CVD, and moderate/severe renal disease were associated with having a major perioperative complication or 30-d readmission. Age was not independently associated with having a surgical complication or readmission. This knowledge is important for preoperative risk stratification of patients when determining whether to electively operate for NPH, can be used to facilitate shared decision making among patients and providers, and suggests that there may be an opportunity to create risk calculators for patient groups undergoing late-in-life elective neurosurgery.
Disclosures
This study was supported by a TL1 Training Grant (to Drs Nadel and Linzey; 1-TL1-TR-002242-01) and an NS-007222 grant (to Dr Wilkinson), both from the National Institutes of Health (NIH). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.
Supplementary Material
Supplemental Digital Content 1. Table. ICD‐9‐CM codes. ICD‐9‐CM codes for post‐surgical complications and Charlson comorbidities.
Supplemental Digital Content 2. Table. Reasons for readmission. Most common reasons for 30‐d readmission and their ICD‐9‐CM codes.
Supplemental Digital Content 3. Table. Risk factors for reoperation within 30 d.
Supplemental Digital Content 4. Table. Bivariate logistic regression models of factors that predict surgical complication or readmission.
REFERENCES
- 1. Adams RD, Fisher CM, Hakim S, Ojemann RG, Sweet WH. Symptomatic occult hydrocephalus with “normal” cerebrospinal-fluid pressure. A treatable syndrome. N Engl J Med. 1965;273(3):117-126. [DOI] [PubMed] [Google Scholar]
- 2. Hakim S, Adams RD. The special clinical problem of symptomatic hydrocephalus with normal cerebrospinal fluid pressure. Observations on cerebrospinal fluid hydrodynamics. J Neurol Sci. 1965;2(4):307-327. [DOI] [PubMed] [Google Scholar]
- 3. Wallenstein MB, McKhann GM. Salomón Hakim and the discovery of normal-pressure hydrocephalus. Neurosurgery. 2010;67(1):155-159; discussion 159. [DOI] [PubMed] [Google Scholar]
- 4. Brean A, Eide PK. Prevalence of probable idiopathic normal pressure hydrocephalus in a Norwegian population. Acta Neurol Scand. 2008;118(1):48-53. [DOI] [PubMed] [Google Scholar]
- 5. Jaraj D, Rabiei K, Marlow T, Jensen C, Skoog I, Wikkelsø C. Prevalence of idiopathic normal-pressure hydrocephalus. Neurology. 2014;82(16):1449-1454. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Martín-Láez R, Caballero-Arzapalo H, Valle-San Román N, López-Menéndez LÁ, Arango-Lasprilla JC, Vázquez-Barquero A. Incidence of idiopathic normal-pressure hydrocephalus in Northern Spain. World Neurosurg. 2016;87:298-310. [DOI] [PubMed] [Google Scholar]
- 7. Tanaka N, Yamaguchi S, Ishikawa H, Ishii H, Meguro K. Prevalence of possible idiopathic normal-pressure hydrocephalus in Japan: The Osaki-Tajiri project. Neuroepidemiology. 2009;32(3):171-175. [DOI] [PubMed] [Google Scholar]
- 8. Thompson SD, Shand Smith JD, Khan AA, Luoma AMV, Toma AK, Watkins LD. Shunting of the over 80s in normal pressure hydrocephalus. Acta Neurochir. 2017;159(6):987-994. [DOI] [PubMed] [Google Scholar]
- 9. Hung AL, Vivas-Buitrago T, Adam A et al.. Ventriculoatrial versus ventriculoperitoneal shunt complications in idiopathic normal pressure hydrocephalus. Clin Neurol Neurosurg. 2017;157:1-6. 10.1016/j.clineuro.2017.03.014. [DOI] [PubMed] [Google Scholar]
- 10. McGovern RA, Kelly KM, Chan AK, Morrissey NJ, McKhann GM. Should ventriculoatrial shunting be the procedure of choice for normal-pressure hydrocephalus? J Neurosurg. 2014;120(6):1458-1464. [DOI] [PubMed] [Google Scholar]
- 11. Poca MA, Solana E, Martínez-Ricarte FR, Romero M, Gándara D, Sahuquillo J. Idiopathic normal pressure hydrocephalus: results of a prospective cohort of 236 shunted patients. Acta Neurochir Suppl. 2012;114:247-253. [DOI] [PubMed] [Google Scholar]
- 12. Schenker P, Stieglitz LH, Sick B, Stienen MN, Regli L, Sarnthein J. Patients with a normal pressure hydrocephalus shunt have fewer complications than do patients with other shunts. World Neurosurg. 2018;110:e249-e257. [DOI] [PubMed] [Google Scholar]
- 13. Chan AK, McGovern RA, Zacharia BE et al.. Inferior short-term safety profile of endoscopic third ventriculostomy compared with ventriculoperitoneal shunt placement for idiopathic normal-pressure hydrocephalus: a population-based study. Neurosurgery. 2013;73(6):951-961. [DOI] [PubMed] [Google Scholar]
- 14. Khan QU, Wharen RE, Grewal SS et al.. Overdrainage shunt complications in idiopathic normal-pressure hydrocephalus and lumbar puncture opening pressure. J Neurosurg. 2013;119(6):1498-1502. [DOI] [PubMed] [Google Scholar]
- 15. Easterlin MC, Chang DG, Talamini M, Chang DC. Older age increases short-term surgical complications after primary knee arthroplasty. Clin Orthop Relat Res. 2013;471(8):2611-2620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Manku K, Bacchetti P, Leung JM. Prognostic significance of postoperative in-hospital complications in elderly patients. I. Long-term survival. Anesth Analg. 2003;96(2):583-589, table of contents. [DOI] [PubMed] [Google Scholar]
- 17. Sieber FE, Barnett SR. Preventing postoperative complications in the elderly. Anesth Clin. 2011;29(1):83-97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Fan W, Sarfarazi M. SAS Solutions to Identifying Hospital Readmissions. 2014. SAS Institute Inc; Paper1622-2014. http://support.sas.com/resources/papers/proceedings14/1622-2014.pdf. Accessed 23 April 2018. [Google Scholar]
- 19. National Cancer Institute Comorbidity Index Overview. https://healthcaredelivery.cancer.gov/seermedicare/considerations/comorbidity.html. Accessed 23 April 2018. [Google Scholar]
- 20. National Cancer Institute Comorbidity SAS Macro, 2014 version https://healthcaredelivery.cancer.gov/seermedicare/considerations/macro-2014.html. Accessed 23 April2018. [Google Scholar]
- 21. Korinek A-M, Fulla-Oller L, Boch A-L, Golmard J-L, Hadiji B, Puybasset L. Morbidity of ventricular cerebrospinal fluid shunt surgery in adults: an 8-year study. Neurosurgery. 2011;68(4):985-995. [DOI] [PubMed] [Google Scholar]
- 22. Rinaldo L, Brown D, Lanzino G, Parney IF. Outcomes following cerebrospinal fluid shunting in high-grade glioma patients. J Neurosurg. 2017;1-13. (doi: 10.3171/2017.6.JNS17859). [DOI] [PubMed] [Google Scholar]
- 23. Guo L, Chen X, Yu B, Shen L, Zhang X. Delayed intracerebral hemorrhage secondary to ventriculoperitoneal shunt: a retrospective study. World Neurosurg. 2017;107:160-167. [DOI] [PubMed] [Google Scholar]
- 24. Hudson JS, Nagahama Y, Nakagawa D et al.. Hemorrhage associated with ventriculoperitoneal shunt placement in aneurysmal subarachnoid hemorrhage patients on a regimen of dual antiplatelet therapy: a retrospective analysis. J Neurosurg. 2017;1-6. (doi: 10.3171/2017.5.JNS17642). [DOI] [PubMed] [Google Scholar]
- 25. Burton JK, Ferguson EEC, Barugh AJ et al.. Predicting discharge to institutional long-term care after stroke: a systematic review and metaanalysis. J Am Geriatr Soc. 2018;66(1):161-169. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Van der Cruyssen K, Vereeck L, Saeys W, Remmen R. Prognostic factors for discharge destination after acute stroke: a comprehensive literature review. Disabil Rehabil. 2015;37(14):1214-1227. [DOI] [PubMed] [Google Scholar]
- 27. Luppa M, Luck T, Weyerer S, König H-H, Brähler E, Riedel-Heller SG. Prediction of institutionalization in the elderly. A systematic review. Age Ageing. 2010;39(1):31-38. [DOI] [PubMed] [Google Scholar]
- 28. Watt J, Tricco AC, Talbot-Hamon C et al.. Identifying older adults at risk of harm following elective surgery: a systematic review and meta-analysis. BMC Med. 2018;16:2 (doi: 10.1186/s12916-017-0986-2). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Kestle JRW. Editorial: Administrative database research. J Neurosurg. 2014;122(2):441-442. [DOI] [PubMed] [Google Scholar]
- 30. Oravec CS, Motiwala M, Reed K et al.. Big data research in neurosurgery: a critical look at this popular new study design. Neurosurgery. 2018;82(5):728-746. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.