Abstract
Background/Objectives:
Our objectives were to, 1) characterize patient/clinical characteristics of adults hospitalized with meningitis, 2) describe meningitis hospitalization outcomes, including 30- and 90- day readmissions, and 3) determine whether clinical, patient or index hospitalization characteristics are associated with readmission/readmission outcomes.
Materials and Methods:
Retrospective study of the 2014 National Readmissions Database. We extracted data on hospitalized adults with a principal diagnosis of meningitis and examined hospitalization outcomes using descriptive statistics. Logistic regression models were built to determine whether characteristics were associated with 30- or 90-day readmission.
Results:
18, 883 adults qualified for 30-day readmission analyses. Meningitis hospitalizations commonly involved adults ages 25–54 who were insured by private carriers. The readmission rate was 7.0% at 30 days, and 11.4% at 90 days. Readmission was associated with greater comorbidity burden (2 conditions: AOR 1.60, 1.24–2.08; 3 conditions: AOR 1.92, 1.43–2.58; 4+ conditions: AOR 2.68, 2.04–3.51 versus 0–1 condition), public insurance (Medicare: AOR 1.85, 1.30–2.62; Medicaid: AOR 1.48, 1.16–1.90 versus private), and medical error (AOR 1.43, 1.07–1.91). Readmissions were most often for meningitis, septicemia, or medical complications.
Conclusions:
Readmission after hospitalization for meningitis is associated with both fixed and modifiable factors. More research is needed to determine which post-meningitis readmissions are preventable.
Keywords: Meningitis, Readmission, Outcomes
Introduction
Meningitis is a central nervous system disorder usually caused by viral, bacterial, or fungal infection.1 Previous research on meningitis outcomes has focused on clinical sequelae,2–7 and several studies have examined mortality outcomes in meningitis2–4,6,8–11; however, there are limited data on other hospitalization outcomes like cost, length of stay, discharge disposition and readmissions. No study has explicitly examined factors associated with meningitis readmissions. This gap in the literature is concerning given that hospitalization data are increasingly being interpreted as indicators of care quality, and used to guide reimbursement strategies in the United States.12
In this study, we examined meningitis hospitalization outcomes in the United States using a national, all-payer dataset. Our study objectives were 1) to characterize patient and clinical characteristics of adults hospitalized with meningitis, 2) to describe meningitis hospitalization outcomes, including 30- and 90- day readmissions, and 3) to determine whether clinical, patient or index hospitalization characteristics are associated with readmission and readmission outcomes.
Materials and Methods
Approvals and Research Protections
The University of Pennsylvania Human Protection Research Organization approved an exemption for the use of Health Care Utilization Project data for research analyses and publication.
Data Source and Study Population
We used the 2014 National Readmissions Database (NRD), a database developed for the Healthcare Cost and Utilization Project (HCUP), for our hospitalization and readmissions analyses. The NRD contains data sampled from approximately fifty percent of hospitalizations in the United States and can be weighted to produce national estimates of hospitalizations and associated outcomes.13 The NRD contains clinical (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] diagnosis and procedure codes, comorbid disease indicators) and nonclinical variables (patient demographics, costs, hospital characteristics).13 Patient identification numbers (created solely for the dataset) allow researchers to track an individual across all hospitalizations during a given year, thus the NRD is the only national source of readmissions data. However, the NRD does not have race data or hospital identifiers that would allow researchers to perform hierarchical analyses (such as those which examine hospital volume or hospital level effects on outcomes).13 We sampled NRD discharges of adults hospitalized with a principal discharge diagnosis of meningitis, as identified by ICD-9 diagnosis codes that have been grouped by HCUP into clinically meaningful categories.14 HCUP Clinical Classification Software classifies individual ICD-9 codes into clinically similar groups.14 We excluded individuals with missing values for any of the variables needed to accomplish the study objectives (i.e., patient and hospital descriptors, clinical and outcome variables). We also excluded individuals who had hospitalizations that disqualified them from readmission analyses- those not residing in the state in which they were initially hospitalized, because we would not have been able to observe local readmissions, and those who died before the 30 or 90 day observation window ended.
Patient, Clinical and Hospital Characteristics
Several variables were extracted or derived from NRD data to be used in descriptive analyses. We extracted sociodemographic information including age, sex, expected payer (uninsured, private insurance, Medicare, Medicaid), and socioeconomic status (indicated by median postal code income quartile) from the initial hospitalization record. Hospital characteristics of interest included hospital bed size (small, medium, large), teaching-population density (metropolitan teaching, metropolitan non-teaching, non-metropolitan/rural) and type of hospital (government, private not-for-profit, or private for-profit).
HCUP clinical classification software was used to identify preexisting comorbid conditions and to calculate an Elixhauser Comorbidity Index Score at the time of initial hospitalization.15 The Elixhauser Comorbidity Index Score was examined as a categorical variable (0–1, 2, 3, 4+ conditions). Mechanical ventilation, seizure and use of continuous electroencephalogram (cEEG) were selected a priori as indicators of a complicated index hospitalization for meningitis; indicator variables for these conditions were created using standard ICD-9 procedure codes. We also extracted data on medical care-related adverse events due to “drugs, medicinal and biological substances causing adverse effects in therapeutic use” (E codes: E9300- E9499) and “misadventures to patients during surgical and medical care” (E codes: E8700-E8799).
Outcomes
Our primary outcomes were 1) index hospitalization disposition, and 2) 30- day and 90-day readmission. Secondary outcomes included 1) readmission discharge disposition, 2) length of stay (LOS), and 3) reason for readmission. Disposition in the NRD was categorized as routine discharge to home, discharge to post-acute care (skilled nursing facility or inpatient rehabilitation), or discharge to home with home health care. The reason for readmission was identified using the principle diagnosis on the readmission record, and classified using HCUP Clinical Classification Software, which groups individual ICD-9 codes into clinically similar groups.14
Thirty-day readmissions were defined as all-cause, non-elective readmission within 30 days of the index meningitis hospitalization discharge date. Individuals who died during the index admission were excluded from all analyses. We also excluded persons with an index hospitalization discharge occurring less than 30 days before December 31, 2014, as the NRD does not allow researchers to track readmissions between calendar years. For instances with more than one readmission, we extracted data only from the first readmission. As a secondary analysis, we examined 90-day readmissions, adjusting the study sample accordingly.
Statistical Analysis
Survey weights were applied to the 2014 NRD to produce national estimates of inpatient care and outcomes for meningitis in the United States. Sociodemographic, clinical and hospital characteristics associated with inpatient care for meningitis were analyzed using descriptive statistics. Weighted, unconditional logistic regression models were built to examine the associations between sociodemographic, clinical, hospital, and index hospitalization characteristics and adjusted odds of 30- and 90-day readmission. We produced rank order lists of the ten most common reasons for readmissions. Statistical analyses were performed using SAS v.9.4 (SAS Institute Inc., Cary, NC, USA).
Results
Study Sample Characteristics
Applying NRD sample weights, we identified 18,883 adults hospitalized for clinical meningitis in the United States who qualified for 30-day readmission analyses. As shown in Table 1, persons ages 25–54 accounted for 58.9% of adult meningitis hospitalizations in our sample. Females were more common in our sample (55.5% versus 44.5%, respectively). Thirteen percent (n=2,476) of adults hospitalized with meningitis were uninsured, and the remainder received benefits from Medicare (17.6%), Medicaid (19.0%), and private insurers (50.3%). Care for meningitis occurred most frequently at large (57.5%), private, not-for-profit (74.3%) and metropolitan teaching hospitals (64.8%).
Table 1.
Index characteristics of meningitis patients eligible for 30 and 90 day readmission analysis, NRD 2014
| Characteristic | Eligible for 30 day readmission analyses N (%) | Eligible for 90 day readmission analyses N (%) |
|---|---|---|
| Age | ||
| 18–24 | 2663 (14.1) | 2196 (14.1) |
| 25–34 | 4385 (23.2) | 3605 (23.1) |
| 35–44 | 3532 (18.7) | 2904 (18.6) |
| 45–54 | 3219 (17.0) | 2705 (17.3) |
| 55–64 | 2469 (13.1) | 2035 (13.0) |
| 65–74 | 1489 (7.9) | 1247 (8.0) |
| 75–84 | 826 (4.4) | 687 (4.4) |
| 85+ | 301 (1.6) | 236 (1.5) |
| Sex | ||
| Male | 8401 (44.5) | 6963 (44.6) |
| Female | 10482 (55.5) | 8651 (55.4) |
| Primary Payer | ||
| Medicare | 3331 (17.6) | 2757 (17.7) |
| Medicaid | 3579 (19.0) | 2936 (18.8) |
| Private insurance | 9498 (50.3) | 7837 (50.2) |
| Other | 2476 (13.1) | 2085 (13.4) |
| Zip–code quartile | ||
| $1 – $37,999 | 4660 (24.7) | 3942 (25.2) |
| $38,000 – $47,999 | 5070 (26.8) | 4213 (27.0) |
| $48,000 – $63,999 | 4728 (25.0) | 3880 (24.8) |
| $64000+ | 4426 (23.4) | 3580 (22.9) |
| Control/ownership of hospital | ||
| Government, nonfederal | 2563 (13.6) | 2143 (13.7) |
| Private, not-for-profit | 14025 (74.3) | 11594 (74.2) |
| Private, for-profit | 2296 (12.2) | 1878 (12.0) |
| Hospital bedsize | ||
| small | 2883 (15.3) | 2340 (15.0) |
| medium | 5141 (27.2) | 4185 (26.8) |
| large | 10860 (57.5) | 9090 (58.2) |
| Teaching status | ||
| metropolitan non-teaching | 5349 (28.3) | 4369 (28.0) |
| metropolitan teaching | 12228 (64.8) | 10166 (65.1) |
| non-metropolitan | 1307 (6.9) | 1080 (6.9) |
HCUP DUA prevents printing cells with N < 10
As displayed in Table 2, most individuals hospitalized for meningitis were otherwise healthy. Fifty-three percent (n=10,022) had no or 1 Elixhauser condition. The most frequently identified chronic conditions were hypertension (31.0%), fluid and electrolyte disorders (28.5%), obesity (12.1%), chronic pulmonary disease (12.0%), uncomplicated diabetes (11.7%), depression (11.2%) and deficiency anemias (11.0%).
Table 2.
Index clinical characteristics of meningitis patients eligible for 30 and 90 day readmission analysis, NRD 2014
| Clinical Characteristic | Eligible for 30 day readmission analyses N (%) | Eligible for 90 day readmission analyses N (%) |
|---|---|---|
| Elixhauser Comorbidity Index Score | ||
| 0–1 condition | 10022 (53.1) | 8307 (53.2) |
| 2 condition | 3327 (17.6) | 2723 (17.4) |
| 3 conditions | 2425 (12.8) | 1991 (12.7) |
| 4+ conditions | 3109 (16.5) | 2595 (16.6) |
| Chronic Conditions | ||
| AIDS | 101 (0.5) | 74 (0.5) |
| Deficiency anemias | 2082(11.0) | 1769(11.3) |
| Rheumatoid arthritis/collagen vascular | ||
| diseases | 606 (3.2) | 511 (3.3) |
| Chronic blood loss anemia | 71 (0.4) | 56 (0.4) |
| Congestive heart failure | 383 (2.0) | 342 (2.2) |
| Chronic pulmonary disease | 2274 (12.0) | 1895 (12.1) |
| Coagulopathy | 835 (4.4) | 679 (4.3) |
| Depression | 2114 (11.2) | 1704 (10.9) |
| Diabetes, uncomplicated | 2206 (11.7) | 1816 (11.6) |
| Diabetes, complicated | 416 (2.2) | 365 (2.3) |
| Drug abuse | 1018 (5.4) | 851 (5.4) |
| Hypertension | 5862 (31.0) | 4846 (31.0) |
| Hypothyroidism | 1522 (8.1) | 1267 (8.1) |
| Liver disease | 416 (2.2) | 345 (2.2) |
| Lymphoma | 167 (0.9) | 131 (0.8) |
| Fluid and electrolyte disorders | 5383 (28.5) | 4461 (28.6) |
| Metastatic cancer | 163 (0.9) | 130 (0.8) |
| Neurological disorders | 84 (0.4) | 70 (0.4) |
| Obesity | 2289 (12.1) | 1872 (12) |
| Paralysis | 316 (1.7) | 276 (1.8) |
| Peripheral vascular disorders | 335 (1.8) | 280 (1.8) |
| Psychoses | 750 (4.0) | 612 (3.9) |
| Pulmonary circulation disorders | 152 (0.8) | 121 (0.8) |
| Renal failure | 773 (4.1) | 655 (4.2) |
| Solid tumor without metastasis | 162 (0.9) | 108 (0.7) |
| Peptic ulcer disease excluding bleeding | * | * |
| Valvular disease | 339 (1.8) | 270 (1.7) |
| Weight loss | 415 (2.2) | 363 (2.3) |
HCUP data use agreement prevents display of table cells with less than 10.
Index admission Outcomes
As shown in Table 3, the majority (51.1%) of patients was discharged between three and six days after admission, and 81.4% of discharges after the index admission were routine discharges to home. Notably, 10.8% (n=2037) of meningitis patients had a medical error code recorded during the index admission; 3.2% (n=612) had medical care errors, and 7.9% (n=1482) had documented injuries from medical drugs. Mechanical ventilation, seizure and continuous EEG use were uncommon, documented in 3.0%, 6.1%, and 0.3% respectively, of hospitalizations.
Table 3.
Outcomes associated with meningitis patients eligible for 30 and 90 day readmission analysis, NRD 2014
| Outcome | Eligible for 30 day readmission analyses N (%) | Eligible for 90 day readmission analyses N (%) |
|---|---|---|
| Length of stay | ||
| 2 days | 5636 (29.8) | 4604 (29.5) |
| 3–6 days | 9654(51.1) | 7989(51.2) |
| >7 days | 3594(19) | 3022 (19.4) |
| Disposition after Index Admission | ||
| routine to home | 15379 (81.4) | 12708 (81.4) |
| transfer to short term hospital | ||
| or other facility | 1435 (7.6) | 1184 (7.6) |
| Home Health Care | 1806 (9.6) | 1498 (9.6) |
| Other | 262 (1.4) | 224 (1.4) |
| Complication | ||
| E code during index Admission (any) | 2037 (10.8) | 1678(10.7) |
| E codes for medical errors/CCS 2616 | 612 (3.2) | 504 (3.2) |
| E codes for drug errors/CCS 2617 | 1482 (7.9) | 1219 (7.8) |
| Specific Complications | ||
| Mechanical Ventilation | 570 (3.0) | 497 (3.2) |
| Seizure | 1157 (6.1) | 978 (6.3) |
| cEEG | 64 (0.3) | 54 (0.3) |
*HCUP data use agreement prevents display of table cells with less than 10.
30-day Readmission
The all- cause 30-day non-elective readmission rate among meningitis inpatients was 7.0%. As shown in Table 4, the readmission rate was lowest among individuals ages 18–24 (4.1%) and highest among individuals ages 85 and older (13.2%). Readmission did not vary by sex. Medicare and Medicaid program participants and individuals from the lowest income neighborhoods had higher than average 30-day readmission rates (13.2%, 8.3% and 8.4%, respectively). Persons requiring advanced post-acute care, such as inpatient post-acute care or home health care had readmission rates nearly twice the average (14.9% and 12.5% respectively).
Table 4.
Factors associated with 30 and 90 day non-elective readmission after hospitalization for meningitis, NRD 2014
| 30 day | 90 day | |||||||
|---|---|---|---|---|---|---|---|---|
| Characteristic | Total population (index admissions) N=18883 | Number readmitted N=1334 | Percent readmitted Overall= 7.0% | Adjusted Odds Ratio of readmission (95% CI) | Total population (index admissions) N=15614 | Number readmitted N=1653 | Percent readmitted Overall =11.4% | Adjusted Odds Ratio of readmission (95% CI) |
| Age | ||||||||
| 18–24 | 2663 | 110 | 4.1 | Ref. | 2196 | 127 | 5.8 | Ref. |
| 25–34 | 4385 | 237 | 5.4 | 1.28 (0.86–1.89) | 3605 | 241 | 6.7 | 1.07 (0.74–1.55) |
| 35–44 | 3532 | 206 | 5.8 | 1.20 (0.81–1.78) | 2904 | 264 | 9.1 | 1.26 (0.89–1.79) |
| 45–54 | 3219 | 241 | 7.5 | 1.28 (0.87–1.89) | 2705 | 306 | 11.3 | 1.29 (0.90–1.84) |
| 55–64 | 2469 | 225 | 9.1 | 1.25 (0.81–1.93) | 2035 | 272 | 13.3 | 1.12 (0.74–1.7) |
| 65–74 | 1489 | 174 | 11.7 | 0.94 (0.57–1.56) | 1247 | 251 | 20.1 | 0.95 (0.60–1.49) |
| 75–84 | 826 | 100 | 12.1 | 0.84 (0.49–1.43) | 687 | 135 | 19.7 | 0.75 (0.46–1.23) |
| 85+ | 301 | 40 | 13.2 | 0.81 (0.40–1.65) | 236 | 58 | 24.4 | 0.89 (0.47–1.7) |
| Sex | ||||||||
| Male | 8401 | 599 | 7.1 | Ref. | 6963 | 767 | 11.0 | Ref. |
| female | 10482 | 735 | 7.0 | 0.99 (0.81–1.22) | 8651 | 886 | 10.2 | 0.93 (0.77–1.12) |
| Payer | ||||||||
| Medicare | 3331 | 441 | 13.2 | 1.85 (1.30–2.62)** | 2757 | 613 | 22.2 | 2.09 (1.53–2.86)** |
| Medicaid | 3579 | 298 | 8.3 | 1.48 (1.16–1.90)** | 2936 | 349 | 11.9 | 1.50 (1.20–1.88)** |
| Private | 9498 | 459 | 4.8 | REF | 7837 | 521 | 6.7 | REF |
| Other | 2476 | 136 | 5.5 | 1.06 (0.74–1.50) | 2085 | 170 | 8.1 | 1.10 (0.80–1.51) |
| Zip–code quartile | ||||||||
| $1 – $37,999 | 4660 | 394 | 8.4 | 1.20 (0.91–1.59) | 3942 | 529 | 13.4 | 1.48 (1.15–1.90)** |
| $38,000 – $47,999 | 5070 | 385 | 7.6 | 1.19 (0.92–1.54) | 4213 | 440 | 10.4 | 1.22 (0.95–1.57) |
| $48,000 – $63,999 | 4728 | 288 | 6.1 | 1.00 (0.75–1.32) | 3880 | 385 | 9.9 | 1.22 (0.93–1.61) |
| $64000+ | 4426 | 268 | 6.1 | Ref. | 3580 | 300 | 8.4 | Ref. |
| Index stay disposition | ||||||||
| Home | 15379 | 847 (5.5) | 5.5 | Ref. | 12708 | 1029 | 8.1 | Ref. |
| Inpatient post-acute care | 1435 | 214 (14.9) | 14.9 | 1.44 (1.06–1.95)** | 1184 | 308 | 26.0 | 1.59 (1.20–2.12)** |
| Home health care | 1806 | 226 (12.5) | 12.5 | 1.64 (1.26–2.13)** | 1498 | 267 | 17.9 | 1.50 (1.16–1.95)** |
| Other/Unknown | 262 | 47 (17.9) | 17.9 | 3.09 (1.82–5.24)** | 224 | 49 | 21.8 | 2.58 (1.51–4.4)** |
| Elixhauser Comorbidity score | ||||||||
| 0–1 conditions | 10022 | 397 | 4.0 | Ref. | 8307 | 431 | 5.2 | Ref. |
| 2 conditions | 3327 | 236 | 7.1 | 1.60 (1.24–2.08)** | 2723 | 246 | 9.0 | 1.51 (1.17–1.96)** |
| 3 conditions | 2425 | 230 | 9.5 | 1.92 (1.43–2.58)** | 1991 | 322 | 16.2 | 2.55 (1.94–3.35)** |
| 4+ conditions | 3109 | 471 | 15.2 | 2.68 (2.04–3.51)** | 2595 | 654 | 25.2 | 3.55 (2.76–4.58)** |
| Length of Index stay | ||||||||
| <3 days | 5636 | 294 | 5.2 | Ref. | 4604 | 320 | 7.0 | Ref. |
| 3–6 days | 9654 | 574 | 5.9 | 0.92 (0.71–1.18) | 7989 | 719 | 9.0 | 1.00 (0.78–1.27) |
| >6 days | 3594 | 466 | 13.0 | 1.30 (0.93–1.80) | 3022 | 614 | 20.3 | 1.34 (0.99–1.82) |
| Teaching status | ||||||||
| metropolitan non-teaching | 5349 | 323 | 6.0 | 0.86 (0.70–1.05) | 4369 | 426 | 9.7 | 0.96 (0.80–1.16) |
| metropolitan teaching | 12228 | 922 | 7.5 | Ref. | 10166 | 1131 | 11.1 | Ref. |
| Non-metropolitan | 1307 | 88 | 6.7 | 0.85 (0.55–1.31) | 1080 | 96 | 8.9 | 0.70 (0.46–1.06) |
| Any E code during index admission | ||||||||
| Absent | 16846 | 1105 | 6.6 | Ref. | 13936 | 1380 | 9.9 | Ref. |
| Present | 2037 | 228 | 11.2 | 1.43 (1.07–1.91)** | 1678 | 273 | 16.3 | 1.33 (1.02–1.74)** |
| E codes for Medical care E8700–E8799/CCS 2616 | ||||||||
| Absent | 18272 | 1258 | 6.9 | Ref. | 15110 | 1566 | 10.4 | Ref. |
| Present | 612 | 76 | 12.4 | 1.61 (0.94–2.74) | 504 | 87 | 17.3 | 1.47 (0.91–2.40) |
| E codes for Adverse Effects of Medical Drugs E9300–E9499/CCS 2617 | ||||||||
| Absent | 17401 | 1173 | 6.7 | Ref. | 14395 | 1454 | 10.1 | Ref. |
| Present | 1482 | 161 | 10.8 | 1.30 (0.97–1.74) | 1219 | 199 | 16.3 | 1.28 (0.97–1.70) |
| Specific Complications (yes vs. no) | ||||||||
| Mechanical ventilation | 570 | 55 | 9.7 | 0.59 (0.33–1.06) | 497 | 100 | 20.1 | 0.79 (0.49–1.29) |
| cEEG | 64 | * | * | 0.35 (0.09–1.44) | 54 | * | * | 0.68 (0.24–1.93) |
| seizure | 1157 | 168 | 14.5 | 1.22 (0.89–1.66) | 978 | 242 | 24.7 | 1.37 (1.06–1.79)** |
Adjusted for sex, ECI, index disposition, LOS, insurance, zip code income quartile, and readmission hospital teaching status.
Indicates statistical significance.
Readmission rates increased with comorbidity burden, from 4.0% (n=397) among persons with 0 or 1 comorbid condition, to 15.2% (n=471) among those with 4 or more comorbid conditions. Readmission rates also increased with increasing index length of stay (Table 4). Eleven percent of persons with an adverse event documented during their index admission were readmitted, versus 6.6% of those without adverse event codes.
Logistic regression models that included variables for patient, clinical, and hospital characteristics found that public insurance program participation (Medicare AOR= Adjusted Odds Ratio 1.85, 95%CI 1.30–2.62 or Medicaid AOR 1.48, 95%CI 1.16–1.90 as compared to private insurance), Elixhauser scores of 4 or more (AOR 2.68, 95%CI 2.04–3.51), 3 (AOR 1.92, 95%CI 1.43–2.58), and 2 (AOR 1.60, 95%CI 1.24–2.08) were associated with greater odds of readmission. Discharge to an inpatient facility (AOR 1.44, 95%CI 1.06–1.95) or home health care (AOR 1.64, 95%CI 1.26–2.13), or, documentation of an adverse event (AOR 1.43, 95% CI 1.07–1.91) was also positively associated with readmission. Patient descriptors, hospital characteristics, and diagnoses indicating more severe clinical course were not significantly associated with readmission in our adjusted models.
90-Day Readmission
The study sample eligible for 90-day readmissions had sociodemographic, clinical and hospital characteristics that were nearly identical to those described for the 30-day sample above (Table 1). The 90-day unplanned readmission rate was 11.4% (n=1,653), among 15,614 eligible patients. As shown in Table 4, 64.2% (n=444) of persons readmitted within 90 days after discharge for meningitis were ages 65 and above. Public (Medicare or Medicaid) insurance, residence in a neighborhood with median income less than $37,999, non- routine discharge, documentation of a medical error, increasing Elixhauser comorbidity scores, and documentation of seizure were all associated with increased odds of readmission within 90 day (Table 4).
Readmission Diagnoses
Table 5 displays the top ten primary readmission diagnoses for both 30 and 90-day readmission analyses. The reasons for readmission were diverse, with no clinical condition observed more frequently than 11.6%. Readmissions were most often for meningitis (11.6% of readmissions at 30 days, 10.3% of readmissions at 90 days), followed by septicemia.
Table 5.
Clinical conditions associated with 30 and 90 day unplanned readmissions, NRD 2014
| 30 day readmission analysis | 90 day readmission analysis | |||
|---|---|---|---|---|
| Rank | Condition | N (%) | Condition | N (%) |
| 1 | Meningitis (except that caused by tuberculosis or sexually transmitted disease) | 154 (11.6) | Meningitis (except that caused by tuberculosis or sexually transmitted disease) | 170 (10.3) |
| 2 | Septicemia (except in labor) | 107 (8.0) | Septicemia (except in labor) | 134 (8.1) |
| 3 | Complications of surgical procedures or medical care | 90 (6.7) | Headache; including migraine | 99 (6.0) |
| 4 | Headache; including migraine | 89 (6.7) | Complications of surgical procedures or medical care | 88 (5.3) |
| 5 | Other nervous system disorders | 61 (4.6) | Epilepsy; convulsions | 67 (4.0) |
| 6 | Encephalitis (except that caused by tuberculosis or sexually transmitted disease) | 57 (4.3) | Other nervous system disorders | 65 (3.9) |
| 7 | Epilepsy; convulsions | 46 (3.5) | Complication of device; implant or graft | 51 (3.1) |
| 8 | Complication of device; implant or graft | 34 (2.6) | Encephalitis (except that caused by tuberculosis or sexually transmitted disease) | 49 (3.0) |
| 9 | Acute cerebrovascular disease | 33 (2.5) | Acute cerebrovascular disease | 49 (3.0) |
| 10 | Acute and unspecified renal failure | 30 (2.2) | Pneumonia (except that caused by tuberculosis or sexually transmitted disease) | 34 (2.1) |
| Total | 1334 (1334/18884 = 7.1% readmission rate) | 1653 (1653/15615 = 10.6% readmission rate) | ||
Rounded to the nearest whole number
Discussion/Conclusion
The results from this national analysis increase our understanding of adult meningitis in the United States by providing key data on patient characteristics, clinical, and hospitalization outcomes, including readmissions. Comorbid disease burden and experiencing a medical adverse event were positively associated with readmission. Readmissions were most often for infection or complications of initial care.
We found a national 30- and 90- day readmission rate of 7.0% and 11.4%, respectively, among adults receiving inpatient meningitis care. A Health Care Cost and Utilization Project (HCUP) statistical brief reported a 30-day all cause readmission rate after meningitis hospitalization of 7.8 percent in 2010.16 A study conducted using the Premier Healthcare Database (PHD) spanning the years 2011–2014 found the 30-day all cause readmission rate for meningitis or encephalitis to be 3.2 percent.17 To understand how our estimates differ from these, it must be understood that these other studies used different study populations, readmission definitions, and time frames. The HCUP study included adults and children. The PHD study estimates were for infants and children only. We have produced estimates for adults, filling a gap in the literature. With respect to readmission definitions, the HCUP study, like our study, captured all readmissions within the same state. However, the PHD study only included readmissions to the same index hospital, which would result in lower estimates. Our analysis was conducted for 2014 while the HCUP and PHD estimates were obtained from 2010 and 2011–2014, respectively. Although the incidence of bacterial meningitis has declined since 1997 due to vaccination,8 it is unclear if this would impact readmissions in any way. Overall thirty-day readmissions also have been decreasing with time.18 Consequently, a future trend analysis should be conducted exploring whether there have been any changes in readmission rates for meningitis.
Our data suggest that 30-day readmission rates for meningitis (7.0%) are lower than estimated for more prevalent infectious diseases including septicemia (2014 rate=18.9%), urinary tract infections (2014 rate=15.7%), and pneumonia (2014 rate=15.5%).19 Readmissions for meningitis in our study were also lower than the overall rate of 30-day all-cause readmissions in 2014 (14.0%).19
Factors related to 30- or 90- day readmission in fully adjusted regression models included comorbid disease burden, initial discharge location, and adverse events. Comorbidities are often incorporated into predictive models of hospital readmission.20 In prior studies of meningitis, age was found to be a prognostic factor for adverse clinical outcomes,2,3,9,10,21 and comorbidities increase with age.22 Additionally, meningitis is particularly challenging to manage in older adults with other chronic conditions.21 In our sample, readmission rates were highest among older adults. However, age was not associated with readmission in the adjusted analyses. These findings suggest that premorbid health state may be a more important driver of readmissions than age.
Surprisingly, those with an index stay discharge to home health care had significantly higher odds of readmission, compared to those discharged home or to inpatient post-acute care. Home health care aims to reduce the risk of readmissions; however, the success of this goal is partially dependent on adequate communication between home health care providers and physicians.23 The needs of sicker patients discharged to home health care after hospitalization for meningitis may be insufficiently met. Further study is needed to determine which patients would more adequately benefit from inpatient post-acute care.
In the 30-day readmission analysis, 10.8% of patients experienced an adverse event. Despite adjusting for covariates, adverse events during the index hospitalization significantly increased the odds for both 30- and 90- day readmissions. And, complications of surgical procedures or medical care were the third and fourth most common reasons for readmission in this study. Medical errors are particularly common in critical care settings, with one prospective cohort study finding a one-year incidence of 20.2%.24 We could not determine with certainty whether patients were admitted to an intensive care unit in this dataset; however, meningitis patients, particularly bacterial meningitis patients, are likely to be critically ill and receive higher-level inpatient care. Diagnostic procedures for meningitis also carry established risks. Headache is a common complication of lumbar puncture and was a common reason for readmission.25–27 Brain herniation and local tissue injury are also possible after lumbar puncture.28 Neuroimaging carries a risk of allergic or toxic reaction to intravenous contrast agents. Antibiotics necessary to treat meningitis may cause liver, kidney injury, or allergic reactions. Clinical monitoring of a meningitis patient requires regular blood sampling. Patient, provider, clinical treatment, environmental, and organizational factors (such as provider experience, medications administered, patient-to-nurse ratio) can lead to medication errors or procedure complications.11 We were unable to examine these factors with this dataset, but future analyses using data containing these variables may identify intervenable processes associated with readmission or unfavorable clinical outcomes. Alternatively, given the low absolute number of readmissions due to complications of medical care, our estimates of medical error related readmissions may be within the acceptable range for care of a condition that requires invasive diagnostic tests and multiple drug exposures.
While meningitis can be severe, our analysis suggests that readmissions may not be dependent on index disease severity, as marked by continuous electroencephalogram monitoring, ventilation, or seizure, but instead on medical-care related factors such as medical procedures, devices, medications, and secondary infections. The only indicator of a complicated index hospitalization that was significantly associated with readmission was seizure, at 90 days. This supports the policy of using readmission rates to measure and compare hospital quality within and between hospitals, since readmission rates do not seem to reflect disease severity or diagnosis in this case. Datasets that support hierarchical analyses (nesting of patients within individual hospitals) are needed to confirm and explore this finding further. However, given the association between comorbidities and readmission as well as insurance type and readmission, it is important that any comparisons are case-mix adjusted.
Meningitis was the most common documented reason for readmission in this national sample. Other infections ranking in the top ten for either the 30-day or 90-day readmissions included septicemia, encephalitis, and pneumonia. With this dataset, we cannot determine if infections are a consequence of a new primary infection (hospital- or community-acquired) or recurrent infection. Recurrent, bacterial meningitis is known to occur in about 5–6% of adults.28 In our study, the proportion of readmissions with recurrent meningitis at 30-days was 11.6%. Baseline health status, particularly the presence of immunocompromising comorbid diseases, is likely an important contributor to post-discharge infection risk. Assuring appropriate empiric and final antibiotic prescribing may also reduce such readmissions.29 Because our national dataset did not include laboratory or imaging data, we were unable to corroborate meningitis etiologies (bacterial, viral, fungal), or extent of disease (meningitis, meningoencephalitis, encephalitis). Consequently, we did not stratify our results by meningitis type. Repeat infections may be preventable, underscoring the importance of further study of infection-related readmissions.
In this study, we have leveraged nationally representative data on inpatient care to produce the first national level data on short term hospitalization outcomes among adults with meningitis in the United States. Our overall findings suggest that readmissions may not be dependent on index disease severity, but instead on patient baseline health state and medical-care related factors. Yet, it is important to emphasize the limitations of our study. Data that is generated for billing or documentation of care or services, as found in billing claims or medical records, is subject to coding error or bias. Administrative datasets such as the NRD may not provide clinical details on disease severity or physical function, or may contain these data in an incomplete or biased fashion. All retrospective studies are subject to these limitations, and methods to increase data comprehensiveness (e.g. patient registries, academic center based studies) directly oppose approaches to produce national estimates or improve external validity. Also, we present the most recent data, which could represent a decline in readmissions in response to changes in clinical guidelines for meningitis prevention and treatment or to readmission reduction strategies and penalties. Future trend analyses will allow stakeholders to place our findings in the appropriate historical context, and guide the development of interventions that may need to target medical care, social support, and/or patient predisposing factors. In spite of these limitations, these data will likely become the benchmark for future public health or health service research evaluations of hospitalization outcomes for meningitis in the adult population.
Highlights:
We characterized patient with meningitis, and 30- and 90- day readmissions.
We found a national 30- and 90- day readmission rate of 7.0% and 11.4%, respectively, among adults receiving inpatient meningitis care.
Readmission was associated with greater comorbidity and readmissions were most often for meningitis, septicemia, or medical complications.
These findings suggest that premorbid health state may be a more important driver of readmissions than age in adults.
These data will likely become the benchmark for future public health or health service research evaluations of hospitalization outcomes for meningitis in the adult population.
Acknowledgments
Thank you notes. Ms. Lia Weil assisted with background literature searches. Mr. Derrick Tam assisted with background literature search, manuscript formatting.
Funding: This work was supported by the Departments of Neurology and of Biostatistics Epidemiology and Informatics at the University of Pennsylvania; and the National Institute of Neurological Disorders and Stroke and the National Institutes of Health [ Grant number NIH F31 NS103445 01A1].
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.
Conflict of interest. D.E. no conflict; T.Z. no conflict; D.T no conflict; J.C. no conflict; D.A. no conflict; A.W. no conflict.
References
- 1.Zueter AM, Zaiter A. Infectious Meningitis. Clinical Microbiology Newsletter. 2015;37:43–51. [Google Scholar]
- 2.van de Beek D, de Gans J, Spanjaard L, Weisfelt M, Reitsma JB, Vermeulen M. Clinical features and prognostic factors in adults with bacterial meningitis. N Engl J Med. 2004;351:1849–1859. [DOI] [PubMed] [Google Scholar]
- 3.Weisfelt M, van de Beek D, Spanjaard L, Reitsma JB, de Gans J. Community-acquired bacterial meningitis in older people. J Am Geriatr Soc. 2006;54:1500–1507. [DOI] [PubMed] [Google Scholar]
- 4.Wang AY, Machicado JD, Khoury NT, Wootton SH, Salazar L, Hasbun R. Community-acquired meningitis in older adults: clinical features, etiology, and prognostic factors. J Am Geriatr Soc. 2014;62:2064–2070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.van de Beek D, Schmand B, de Gans J, et al. Cognitive impairment in adults with good recovery after bacterial meningitis. J Infect Dis. 2002;186:1047–1052. [DOI] [PubMed] [Google Scholar]
- 6.Aronin SI, Peduzzi P, Quagliarello VJ. Community-acquired bacterial meningitis: risk stratification for adverse clinical outcome and effect of antibiotic timing. Ann Intern Med. 1998;129:862–869. [DOI] [PubMed] [Google Scholar]
- 7.Auburtin M, Porcher R, Bruneel F, et al. Pneumococcal meningitis in the intensive care unit: prognostic factors of clinical outcome in a series of 80 cases. Am J Respir Crit Care Med. 2002;165:713–717. [DOI] [PubMed] [Google Scholar]
- 8.Castelblanco RL, Lee M, Hasbun R. Epidemiology of bacterial meningitis in the USA from 1997 to 2010: a population-based observational study. Lancet Infect Dis. 2014;14:813–819. [DOI] [PubMed] [Google Scholar]
- 9.Durand ML, Calderwood SB, Weber DJ, et al. Acute bacterial meningitis in adults. A review of 493 episodes. N Engl J Med. 1993;328:21–28. [DOI] [PubMed] [Google Scholar]
- 10.Sigurdardottir B, Bjornsson OM, Jonsdottir KE, Erlendsdottir H, Gudmundsson S. Acute bacterial meningitis in adults. A 20-year overview. Arch Intern Med. 1997;157:425–430. [DOI] [PubMed] [Google Scholar]
- 11.Hasbun R, Rosenthal N, Balada-Llasat JM, et al. Epidemiology of Meningitis and Encephalitis in the United States, 2011–2014. Clin Infect Dis. 2017;65:359–363. [DOI] [PubMed] [Google Scholar]
- 12.Readmissions Reduction Program (HRRP). U.S. Centers for Medicare & Medicaid Services; Accessed at: https://www.cms.gov/medicare/medicare-fee-for-service-payment/acuteinpatientpps/readmissions-reduction-program.html.
- 13.HCUP Nationwide Readmissions Database (NRD). Healthcare Cost and Utilization Project (HCUP). 2013. Agency for Healthcare Research and Quality, Rockville, MD. www.hcup-us.ahrq.gov/nrdoverview.jsp. [Google Scholar]
- 14.HCUP CLinical Classification Software (CCS). Healthcare Cost and Utilization Project (HCUP). 2013. Agency for Healthcare Research and Quality, Rockville, MD. http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp. [Google Scholar]
- 15.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36:8–27. [DOI] [PubMed] [Google Scholar]
- 16.Elixhauser A, Steiner C. Readmissions to U.S. Hospitals by Diagnosis, 2010: Statistical Brief #153. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville (MD): Agency for Healthcare Research and Quality (US); 2006. [PubMed] [Google Scholar]
- 17.Hasbun R, Wootton SH, Rosenthal N, et al. Epidemiology of Meningitis and Encephalitis in Infants and Children in the United States, 2011–2014. Pediatr Infect Dis J. 2019;38:37–41. [DOI] [PubMed] [Google Scholar]
- 18.Fingar K, Washington R. Trends in Hospital Readmissions for Four High-Volume Conditions, 2009–2013: Statistical Brief #196. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville (MD): Agency for Healthcare Research and Quality (US); 2006. [PubMed] [Google Scholar]
- 19.Fingar KR, Barrett ML, Jiang HJ. A Comparison of All-Cause 7-Day and 30-Day Readmissions, 2014: Statistical Brief #230. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville (MD): Agency for Healthcare Research and Quality (US); 2006. [PubMed] [Google Scholar]
- 20.Kansagara D, Englander H, Salanitro A, et al. Risk prediction models for hospital readmission: a systematic review. JAMA. 2011;306:1688–1698. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Domingo P, Pomar V, de Benito N, Coll P. The spectrum of acute bacterial meningitis in elderly patients. BMC Infect Dis. 2013;13:108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ashman JJ, Beresovsky V. Multiple chronic conditions among US adults who visited physician offices: data from the National Ambulatory Medical Care Survey, 2009. Prev Chronic Dis. 2013;10:E64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Pesko MF, Gerber LM, Peng TR, Press MJ. Home Health Care: Nurse-Physician Communication, Patient Severity, and Hospital Readmission. Health Serv Res. 2018;53:1008–1024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Rothschild JM, Landrigan CP, Cronin JW, et al. The Critical Care Safety Study: The incidence and nature of adverse events and serious medical errors in intensive care. Crit Care Med. 2005;33:1694–1700. [DOI] [PubMed] [Google Scholar]
- 25.Ahmed SV, Jayawarna C, Jude E. Post lumbar puncture headache: diagnosis and management. Postgrad Med J. 2006;82:713–716. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Leibold RA, Yealy DM, Coppola M, Cantees KK. Post-dural-puncture headache: characteristics, management, and prevention. Ann Emerg Med. 1993;22:1863–1870. [DOI] [PubMed] [Google Scholar]
- 27.Kuntz KM, Kokmen E, Stevens JC, Miller P, Offord KP, Ho MM. Post-lumbar puncture headaches: experience in 501 consecutive procedures. Neurology. 1992;42:1884–1887. [DOI] [PubMed] [Google Scholar]
- 28.Tebruegge M, Curtis N. Epidemiology, etiology, pathogenesis, and diagnosis of recurrent bacterial meningitis. Clin Microbiol Rev. 2008;21:519–537. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Tunkel Allan R., Hartman Barry J., Kaplan Sheldon L., et al. Practice Guidelines for the Management of Bacterial Meningitis. Clinical Infectious Diseases. Epub 2004.:1267. [DOI] [PubMed] [Google Scholar]