Network analysis: a novel method for mapping neonatal acute transport patterns in California.

OBJECTIVE: The objectives of this study are to use network analysis to describe the pattern of neonatal transfers in California, to compare empirical sub-networks with established referral regions and to determine factors associated with transport outside the originating sub-network. STUDY DESIGN: This cross-sectional database study included 6546 infants <28 days old transported within California in 2012. After generating a graph representing acute transfers between hospitals (n=6696), we used community detection techniques to identify more tightly connected sub-networks. These empirically derived sub-networks were compared with state-defined regional referral networks. Reasons for transfer between empirical sub-networks were assessed using logistic regression. RESULTS: Empirical sub-networks showed significant overlap with regulatory regions (P<0.001). Transfer outside the empirical sub-network was associated with major congenital anomalies (P<0.001), need for surgery (P=0.01) and insurance as the reason for transfer (P<0.001). CONCLUSION: Network analysis accurately reflected empirical neonatal transfer patterns, potentially facilitating quantitative, rather than qualitative, analysis of regionalized health care delivery systems.

Women's prepregnancy underweight as a risk factor for preterm birth: a retrospective study.

OBJECTIVE: To investigate the distribution of known factors for preterm birth (PTB) by severity of maternal underweight; to investigate the risk-adjusted relation between severity of underweight and PTB, and to assess whether the relation differed by gestational age. DESIGN: Retrospective cohort study. SETTING: State of California, USA. METHODS: Maternally linked hospital and birth certificate records of 950 356 California deliveries in 2007-2010 were analysed. Singleton live births of women whose prepregnancy body mass index (BMI) was underweight (<18.5 kg/m2 ) or normal (18.50-24.99 kg/m2 ) were analysed. Underweight BMI was further categorised as: severe (<16.00), moderate (16.00-16.99) or mild (17.00-18.49). PTB was grouped as 22-27, 28-31, 32-36 or <37 weeks (compared with 37-41 weeks). Adjusted multivariable Poisson regression modeling was used to estimate relative risk for PTB. MAIN OUTCOME MEASURES: Risk of PTB. RESULTS: About 72 686 (7.6%) women were underweight. Increasing severity of underweight was associated with increasing percent PTB: 7.8% (n = 4421) in mild, 9.0% (n = 1001) in moderate and 10.2% (475) in severe underweight. The adjusted relative risk of PTB also significantly increased: adjusted relative risk (aRR) = 1.22 (95% CI 1.19-1.26) in mild, aRR = 1.41 (95% CI 1.32-1.50) in moderate and aRR = 1.61 (95% CI 1.47-1.76) in severe underweight. These findings were similar in spontaneous PTB, medically indicated PTB, and the gestational age groupings. CONCLUSION: Increasing severity of maternal prepregnancy underweight BMI was associated with increasing risk-adjusted PTB at <37 weeks. This increasing risk was of similar magnitude in spontaneous and medically indicated births and in preterm delivery at 28-31 and at 32-36 weeks of gestation. TWEETABLE ABSTRACT: Increasing severity of maternal underweight BMI was associated with increasing risk of preterm birth.

Examining the effect of hospital-level factors on mortality of very low birth weight infants using multilevel modeling.

OBJECTIVE: The objective of this study was to examine the effect of hospital-level factors on mortality of very low birth weight infants using multilevel modeling. STUDY DESIGN: This is a secondary data analysis of California maternal-infant hospital discharge data from 1997 to 2002. The study population was limited to singleton, non-anomalous, very low birth weight infants, who delivered in hospitals providing neonatal intensive care services (level-2 and higher). Hierarchical generalized linear modeling, also known as multilevel modeling, was used to adjust for individual-level confounders. RESULT: In a multilevel model, increasing hospital volume of very low birth weight deliveries was associated with lower odds of very low birth weight mortality. Characteristics of a particular hospital's obstetrical and neonatal services (the presence of residency and fellowship training programs and the availability of perinatal and neonatal services) had no independent effect. CONCLUSION: Using multilevel modeling, hospital volume of very low birth weight deliveries appears to be the primary driver of reduced mortality among very low birth weight infants.

Predictors of medical service utilization among individuals with co-occurring HIV infection and substance abuse disorders.

This study examined factors affecting medical service use among HIV-infected persons with a substance abuse disorder. The sample comprised 190 participants enrolled in a randomized trial of a case management intervention. Participants were interviewed about their backgrounds, housing status, income, alcohol and drug use problems, health status and depressive symptoms at study entry. Electronic medical records were used to assess medical service use. Poisson regression models were tested to determine the effects of need, enabling and predisposing factors on the dependent variables of emergency department visits, inpatient admissions and ambulatory care visits. During a two-year period, 71% were treated in the emergency department, 64% had been hospitalized and the sample averaged 12.9 ambulatory care visits. Homelessness was associated with higher utilization of emergency department and inpatient services; drug use severity was associated with higher inpatient and ambulatory care service use; and alcohol use severity was associated with greater use of emergency medical services. Homelessness and substance abuse exacerbate the health care needs of HIV-infected persons and result in increased use of emergency department and inpatient services. Interventions are needed that target HIV-infected persons with substance abuse disorders, particularly those that increase entry and retention in outpatient health care and thus decrease reliance on acute hospital-based services.

Is travel distance a barrier to veterans' use of VA hospitals for medical surgical care?

Lengthy travel distances may explain why relatively few veterans in the United States use VA hospitals for inpatient medical/surgical care. We used two approaches to distinguish the effect of distance on VA use from other factors such as access to alternatives and veterans' characteristics. The first approach describes how disparities in travel distance to the VA are related to other characteristics of geographic areas. The second approach involved a multivariate analysis of VA use in postal zip code areas (ZCAs). We used several sources of data to estimate the number of veterans who had priority access to the VA so that use rates could be estimated. Access to hospitals was characterized by estimated travel distance to inpatient providers that typically serve each ZCA. The results demonstrate that travel distance to the VA is variable, with veterans in rural areas traveling much farther for VA care than veterans in areas of high population density. However, Medicare recipients also travel farther in areas of low population density. In some areas veterans must travel lengthy distances for VA care because VA hospitals which were built over the past few decades are not located close to areas in which veterans reside in the 1990s. The disparities in travel distance suggest inequitable access to the VA. Use of the VA decreases with increases in travel distance only up to about 15 miles, after which use is relatively insensitive to further increases in distance. The multivariate analyses indicate that those over 65 are less sensitive to distance than younger veterans, even though those over 65 are Medicare eligible and therefore have inexpensive access to alternatives. The results suggest that proximity to a VA hospital is only one of many factors determining VA use. Further research is indicated to develop an appropriate response to the needs of the small but apparently dedicated group of VA users who are traveling very long distances to obtain VA care.

Short-term health and economic benefits of smoking cessation: low birth weight.

OBJECTIVES: To estimate excess direct medical costs of low birth weight from maternal smoking and short-term cost savings from smoking cessation programs before or during the first trimester of pregnancy. METHODS: Simulations using data on neonatal costs per live birth. Outcome measures are mean US excess direct medical cost per live birth, total excess direct medical cost, reductions in low birth weight, and savings in medical costs from an annual 1 percentage point drop in smoking prevalence among pregnant women. RESULTS: Mean average excess direct medical cost per live birth for each pregnant smoker (in 1995 dollars) was $511; total cost was $263 million. An annual drop of 1 percentage point in smoking prevalence would prevent 1300 low birth weight live births and save $21 million in direct medical costs in the first year of the program; it would prevent 57,200 low birth weight infants and save $572 million in direct medical costs in 7 years. CONCLUSIONS: Smoking cessation before the end of the first trimester produces significant cost savings from the prevention of low birth weight.

The hospital cost of congenital syphilis.

OBJECTIVE: To determine the hospital cost of caring for newborn infants with congenital syphilis. STUDY POPULATION: All live-born singleton neonates with birth weight greater than 500 gm at an inner-city municipal hospital in New York City in 1989. METHODS: We compared the characteristics of 114 infants with case-compatible congenital syphilis with those of 2906 infants without syphilis. Cost estimates were based on New York State newborn diagnosis-related groups (DRG) reimbursements adjusted for length of stay, birth weight, preterm delivery, and selected maternal risk factors, including infection with the human immunodeficiency virus, cocaine use during pregnancy, and history of injected drug use. RESULTS: For infants with congenital syphilis, the unadjusted mean cost ($11,031) and the median cost ($4961) were more than three times larger than those for infants without syphilis (p < 0.01). After adjustment, congenital syphilis was associated with an additional length of hospitalization of 7 1/2 days and an additional cost of $4690 (both p < 0.01) above mean study population values (7.13 days, $3473). CONCLUSIONS: Based on the number of reported cases (1991 to 1994), the average annual national cost of treating infants with congenital syphilis is approximately $18.4 million (1995 dollars). This estimate provides a benchmark to assess the cost-effectiveness of strategies to prevent, diagnose, and treat the disease.

Does case mix matter for substance abuse treatment? A comparison of observed and case mix-adjusted readmission rates for inpatient substance abuse treatment in the Department of Veterans Affairs.

OBJECTIVE: To develop a case mix model for inpatient substance abuse treatment to assess the effect of case mix on readmission across Veterans Affairs Medical Centers (VAMCs). DATA SOURCES/STUDY SETTING: The computerized patient records from the 116 VAMCs with inpatient substance abuse treatment programs between 1987 and 1992. STUDY DESIGN: Logistic regression was used on patient data to model the effect of demographic, psychiatric, medical, and substance abuse factors on readmission to VAMCs for substance abuse treatment within six months of discharge. The model predictions were aggregated for each VAMC to produce an expected number of readmissions. The observed number of readmissions for each VAMC was divided by its expected number to create a measure of facility performance. Confidence intervals and rankings were used to examine how case mix adjustment changed relative performance among VAMCs. DATA COLLECTION/EXTRACTION METHODS: Ward where care was provided and ICD-9-CM diagnosis codes were used to identify patients receiving treatment for substance abuse (N = 313,886). PRINCIPAL FINDINGS: The case mix model explains 36 percent of the observed facility level variation in readmission. Over half of the VAMCs had numbers of readmissions that were significantly different than expected. There were also noticeable differences between the rankings based on actual and case mix-adjusted readmissions. CONCLUSIONS: Secondary data can be used to build a reasonably stable case mix model for substance abuse treatment that will identify meaningful variation across facilities. Further, case mix has a large effect on facility level readmission rates for substance abuse treatment. Uncontrolled facility comparisons can be misleading. Case mix models are potentially useful for quality assurance efforts.

The effects of patient volume and level of care at the hospital of birth on neonatal mortality.

OBJECTIVE: To examine the effects of neonatal intensive care unit (NICU) patient volume and the level of NICU care available at the hospital of birth on neonatal mortality. DESIGN: Birth certificate data linked to infant death certificates and to infant discharge abstracts were used in a logistic regression model to control for differences in each patient's clinical and demographic risks. Hospitals were classified by the level of NICU care available (no NICU: level I; intermediate NICU: level II; expanded intermediate NICU: level II+: tertiary NICU: level III) and by the average patient census in the NICU. SETTING: All nonfederal hospitals in California with maternity services. PATIENTS: All births in nonfederal hospitals in California in 1990 (N=594104), 473209 (singletons only) of which were successfully linked with discharge abstracts. Of these infants, 53229 were classified as likely NICU admissions. MAIN OUTCOME MEASURES: Death within the first 28 days of life, or within the first year of life, if continuously hospitalized. RESULTS: Patient volume and level of NICU care at the hospital of birth both had significant effects on mortality. Compared with hospitals without an NICU, infants born in a hospital with a level III NICU with an average NICU census of at least 15 patients per day had significantly lower risk-adjusted neonatal mortality (odds ratio, 0.62; 95% confidence interval, 0.47-0.82; P=.002). Risk-adjusted neonatal mortality for infants born in smaller level III NICUs, and in level II+ and level II NICUs, regardless of size, was not significantly different from hospitals without an NICU, and was significantly higher than hospitals with large level III NICUS. CONCLUSIONS: Risk-adjusted neonatal mortality was significantly lower for births that occurred in hospitals with large (average census, >15 patients per day) level III NICUs. Despite the differences in outcomes, costs for the birth of infants born at hospitals with large level III NICUs were not more than those for infants born at other hospitals with NICUs. Concentration of high-risk deliveries in urban areas in a smaller number of hospitals that could provide level III NICU care has the potential to decrease neonatal mortality without increasing costs.

Correlation of travel time on roads versus straight line distance.

Studies of hospital demand and choice of hospital have used straight line distance from a patient's home to hospitals as a measure of geographic access, but there is the potential for bias if straight line distance does not accurately reflect travel time. Travel times for unimpeded travel between major intersections in upstate New York were compared with distances between these points. The correlation between distance and travel time was 0.987 for all observations and 0.826 for distances less than 15 miles. These very high correlations indicate that straight line distance is a reasonable proxy for travel time in most hospital demand or choice models, especially those with large numbers of hospitals. The authors' outlier analyses show some exceptions, however, so this relationship may not hold for studies focusing on specific hospitals, very small numbers of hospitals, or studies in dense urban areas with high congestion and reliance on surface streets.

Inpatient treatment for substance abuse patients with psychiatric disorders: a national study of determinants of readmission.

This study examined the patient case mix and program determinants of 6-month readmission rates and early treatment dropout for 7,711 VA inpatients with both substance abuse and major psychiatric disorders treated in one of 104 substance abuse programs. Patients were treated in one of three types of inpatient programs: explicitly designed dual diagnosis specialty programs, substance abuse programs with a dual diagnosis psychotherapy group or standard substance abuse programs. Dual diagnosis specialty programs differed from regular substance abuse programs in that they had a more severe case mix, a higher 180-day readmission rate, greater dual diagnosis treatment orientation, used more psychotropic medication, had longer lengths of stay, had greater tolerance of relapse and medication noncompliance, and a higher rate of psychiatric aftercare in the 30 days after discharged. Programs with less severe case mix, longer intended and actual length of stay, lower 7-day dropout rates, greater tolerance of problem behavior, 12-step groups, and higher immediate postdischarge utilization of outpatient mental health treatment lower 180-day readmission rates. Programs with less severe patient case mix, more use of psychotropic medications but less of methadone and antabuse, less varied and diverse treatment activities, and low use of patient-led groups had lower dropout rates.

Determinants of readmission following inpatient substance abuse treatment: a national study of VA programs.

This study examines program determinants of one aspect of VA inpatient substance abuse treatment program performance. Performance was measured by the ratio of a program's readmission rate to the expected rate for programs with similar patients. Six-month readmission rates in 101 VA treatment programs were analyzed. Preliminary analyses indicated that patient differences across programs accounted for 36% of the variance in readmission rates. Program differences accounted for 47% of the variance in case-mix-adjusted readmission rate. Among program factors selected through a literature review, better than expected readmission performance was associated with having fewer early discharges, a longer intended treatment duration, more patient participation in aftercare, more family or friend assessment interviews, and treating more patients on a compulsory basis. Performance was not related to stress management training, patient attendance at more self-help meetings during treatment, staff characteristics, or average staff costs per patient day. The findings indicate that treatment retention, duration, and increased aftercare may be targeted to reduce high readmission rates. Last, there were only small differences in the model over 30, 60, 90, and 365 day follow-up intervals, suggesting substantial stability of the findings.

Prenatal care needs assessment comparing service use and outcomes in Fresno, CA.

The authors performed a prenatal care needs assessment for Fresno County, CA, using data from a sample of 11,878 birth certificates for the county for 1989. Birth records, patterns of prenatal care utilization, and low birth weight outcomes in the county were compared with those in a random sample of 11,826 certificates derived from births in the remainder of the State. Bivariate techniques were used in calculating care utilization rates. Multivariate logistic regression analysis was used in associating rates of prenatal care visits and gestational month of initiation of prenatal care with low weight birth outcomes. County women entered prenatal care as early as women in the remainder of the State, but did not return as often for prenatal care visits. Their rate of return for followup visits was 29.9 percent, compared with 24.8 percent for women in all other counties (P < 0.001). County women with the lowest rates of visits had 1.4 to 1.9 times the risk of having a low weight birth than other county women with higher rates of visits, and a significantly higher risk than for women of all other counties. An intensive visit schedule for high-risk care was provided 28.9 percent of county women, compared with 33.0 percent of women in all other counties (P < 0.001). County women who received a high-risk intensive visit schedule were 2.5 times more likely to have a low weight birth than county women who did not receive the schedule. For all other women in the State, the comparable risk was 2.1 times. Improvements in the number and content of prenatal care visits were shown to have a high likelihood of substantially improving birth weight outcomes for pregnancies among Fresno County women.

Cost effects of surfactant therapy for neonatal respiratory distress syndrome.

OBJECTIVE: To examine the cost effects of a single dose (5 ml/kg) of a protein-free synthetic surfactant (Exosurf) as therapy for neonatal respiratory distress syndrome, for both rescue and prophylactic therapy. RESEARCH DESIGN: Nonblinded, randomized clinical trials of both rescue and prophylactic therapy. Regression analyses were used to control for the independent effects of sex, multiple birth, delivery method, birth weight, and surfactant therapy. SETTING: The prophylactic trial was conducted at a university medical center only; the rescue trial also included a tertiary community hospital. PATIENTS: Prophylaxis was administered immediately after birth to 36 infants (38 control subjects) with birth weights between 700 and 1350 gm. Rescue therapy was administered at 4 to 24 hours of age to 53 infants (51 control subjects) with established respiratory distress syndrome and birth weights > or = 650 gm (no upper limit). Infants in the prophylactic trial were not eligible for the rescue trial. RESULTS: For the rescue trial, there was a $16,600 reduction in average hospital costs (p = 0.18), which was larger than the cost of the surfactant ($450 to $900), yielding a probable net savings. For the prophylactic trial, hospital costs were larger for treated infants versus control subjects who weighed less than about 1100 gm at birth and lower for treated infants versus control subjects who weighed more than 1100 gm at birth (p < 0.05). For the prophylactic sample, the result was an average cost per life saved of $71,500. CONCLUSIONS: Single-dose rescue surfactant therapy is probably a cost-effective therapy because it produced a lower mortality rate for the same (and probably lower) expenditure. Single-dose prophylactic therapy for smaller infants (< or = 1350 gm) appeared to yield a reduction in mortality rate for a small additional cost. The use of multiple-dose therapy in infants who do not respond to initial therapy may alter the effects described above to either increase or decrease the observed cost-effectiveness of surfactant therapy. Regardless, surfactant therapy will remain a cost-effective method of reducing mortality rates, relative to other commonly used health care interventions.

A variable-radius measure of local hospital market structure.

OBJECTIVE: To provide a radius measure of the structure of local hospital markets that varies with hospital characteristics and is available for all hospitals in the United States. DATA SOURCES: 1982 American Hospital Association (AHA) Survey of Hospitals, 1982 Area Resource File (ARF), and 1983 California Office of Statewide Health Planning and Development (OSHPD) discharge abstracts. STUDY DESIGN: The OSHPD data were used to measure the radii necessary to capture 75 percent and 90 percent of each hospital's admissions. These radii were used as the dependent variables in regression models in which the independent variables were from the AHA and ARF. To estimate predicted market radii, the estimated parameters from the California models were applied to all nonfederal, short-term, general hospitals in the continental United States. These radii were used to define each hospital's service area, and all other hospitals within the calculated radii were considered potential competitors. Using this definition, we calculated two measures of local market structure: the number of other hospitals within the radius and a Herfindahl-Hirschman Index based on the distribution of hospital bed shares in the market. DATA EXTRACTION METHODS: These measures were calculated for all nonfederal, short-term, acute care hospitals in the continental United States for whom complete data were available (N = 4,884). CONCLUSIONS: These measures are available from the authors on computer-readable diskette, matched to hospital identifiers.

Choice of hospital for delivery: a comparison of high-risk and low-risk women.

OBJECTIVE: This article tests whether or not the factors that affect hospital choice differ for selected subgroups of the population. DATA SOURCES: 1985 California Office of Statewide Health Planning and Development (OSHPD) discharge abstracts and hospital financial data were used. STUDY DESIGN: Models for hospital choice were estimated using McFadden's conditional logit model. Separate models were estimated for high-risk and low-risk patients, and for high-risk and low-risk women covered either by private insurance or by California Medicaid. The model included independent variables to control for quality, price, ownership, and distance to the hospital. DATA EXTRACTION: Data covered all maternal deliveries in the San Francisco Bay Area in 1985 (N = 61,436). ICD-9 codes were used to classify patients as high-risk or low-risk. The expected payment code on the discharge abstract was used to identify insurance status. PRINCIPAL FINDINGS: The results strongly reject the hypothesis that high-risk and low-risk women have the same choice process. Hospital quality tended to be more important for high-risk than low-risk women. These results also reject the hypothesis that factors influencing choice of hospital are the same for women covered by private insurance as for those covered by Medicaid. Further, high-risk women covered by Medicaid were less likely than high-risk women covered by private insurance to deliver in hospitals with newborn intensive care units. CONCLUSIONS: The results show that the choice factors vary across several broadly defined subgroups of patients with a specific condition. Thus, estimates aggregating all patients may be misleading. Specifically, such estimates will understate actual patient response to quality of care indicators, since patient sensitivity to quality of care varies with the patients' risk status.

Birth weight and illness severity: independent predictors of neonatal mortality.

BACKGROUND: Low birth weight is a major determinant of neonatal mortality. Yet birth weight, even in conjunction with other demographic markers, is inadequate to explain the large variations in neonatal mortality between intensive care units. This variation probably reflects differences in admission severity. The authors have recently developed the Score for Neonatal Acute Physiology (SNAP), an illness severity index specific for neonatal intensive care, and demonstrated illness severity to be a major determinant of neonatal mortality. OBJECTIVE: To define the relative contributions of birth weight and illness severity to the risk of neonatal mortality and to identify other significant independent risk factors. METHODS: Logistic regression was used to analyze data from a cohort of 1621 consecutive admissions to three neonatal intensive care units (92 deaths), to test six alternative predictive models. The best logistic model was then used to develop a simple additive clinical score, the SNAP Perinatal Extension (SNAP-PE). RESULTS: These analyses demonstrated that birth weight and illness severity are powerful independent predictors across a broad range of birth weights and that their effects are additive. Below 750 g, there is an interaction between birth weight and SNAP. Other factors that showed independent predictive power were low Apgar score at 5 minutes and small size for gestational age. Separate derivation and test samples were used to demonstrate that the SNAP-PE is comparable to the best logistic model and has a sensitivity and specificity superior to either birth weight or SNAP alone (receiver-operator characteristic area .92 +/- .02) as well as excellent goodness of fit. CONCLUSION: This simplified clinical score provides accurate mortality risk estimates for application in a broad array of clinical and research settings.

Back transporting infants from neonatal intensive care units to community hospitals for recovery care: effect on total hospital charges.

Many neonates are referred to neonatal intensive care units (NICUs) for specialized care far from their parents' residence. This distance can add to the stress of the parents and reduce the contact of the parents with their newborn. Small studies have found that back transporting these neonates to hospitals closer to their homes is safe and cost-effective. Despite these findings, the reluctance of many insurers to pay for back transports prevents or delays many back transports. Insurers may not consider the findings of the previous studies to be conclusive, given that the comparisons were between small numbers of neonates back transported and neonates who remained in tertiary care, and the potential for differences in severity of illness between the groups is significant. In this study the effect on hospital charges of back transports was examined by comparing the charges for care in community hospitals with what these charges would have been in a tertiary care center. The advantage of this method is that it avoids case-mix differences between the groups and thus minimizes the potential for small-sample bias. Data were collected for all back transports from a NICU to non-tertiary care centers (n = 90) for a 9-month period. We were able to obtain the itemized bills for the care at community hospitals for 42 of these patients. Each bill was recalculated using the charges for the NICU to determine potential for savings. The average charges for recovery care were about $6200 lower at the community hospital than they would have been at the NICU.(ABSTRACT TRUNCATED AT 250 WORDS)