 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
,||,¶
* Department of Health Care Policy, Harvard Medical School, Boston, Massachusetts
National Bureau of Economic Research, Cambridge, Massachusetts
Center for Health Policy & Clinical Effectiveness
|| Division of Neonatology, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio
¶ Institute of Health Policy and Health Services Research, University of Cincinnati, Cincinnati, Ohio
# Center for Child Health Care Studies, Department of Ambulatory Care and Prevention, Harvard Pilgrim Health Care and Harvard Medical School, Boston, Massachusetts
** Division of General Pediatrics, Children's Hospital Boston, Boston, Massachuestts
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Methods. A retrospective analysis was conducted of Ohio Medicaid claims data linked with birth certificate data for the period 1991-1998. The impact of the legislation was evaluated using interrupted time-series analysis of health-related utilization. The effects of early follow-up visits for vaginally delivered newborns with short stays were analyzed using the day of the week on which the birth occurred (eg, Monday, Tuesday) as an instrumental variable to account for potential confounding. A total of 155 352 full-term newborns who were born to mothers who receive Medicaid were studied. The main outcomes measured were rehospitalizations, emergency department (ED) visits, and diagnoses of dehydration and infection within 10 and 21 days of birth.
Results. Few outcomes exhibited significant changes after legislation mandating coverage of minimum postnatal hospital stays. Rates of rehospitalization for jaundice within 10 days of birth fell from 0.78% to 0.47% in the year after legislation was introduced but leveled off after the legislation took effect. Rates of ED visits within 21 days increased from 6.0% to 10.4% during periods of increasing short stay but fell to 8.0% during the year after introduction of the legislation and leveled off when the legislation took effect. Rates of all-cause rehospitalization, dehydration, and infection diagnoses showed no consistent relationship to Ohio's legislation. Using instrumental variable analysis, newborns who received early follow-up visits were significantly less likely to have rehospitalizations within the first 10 days of life than those who did not.
Conclusions. In this state Medicaid population, legislation mandating coverage of minimum postnatal hospital stays was associated with reductions in the rates of rehospitalization for jaundice and ED visits. For newborns with short stays, early follow-up visits may reduce rehospitalizations in the early postpartum period.
Key Words: length of stay • newborn • legislation • follow-up • Medicaid
Abbreviations: ED, emergency department • AAP, American Academy of Pediatrics • IV, instrumental variable
Early newborn discharge, the practice of discharging newborns and mothers from the hospital at <48 hours after delivery, resulted in 1 of the most widely discussed public policy changes affecting pediatrics in the past decade. Trends toward earlier hospital discharge of healthy infants began during the 1970s and accelerated during the early 1990s amid increasing financial pressure on insurers and hospitals.1,2 In the mid-1990s, case reports and professional organizations raised questions about the safety of early newborn discharge. By the late 1990s, Congress and 43 states had passed laws requiring that insurers cover hospital stays of at least 48 hours after vaginal delivery.1
Despite the widespread discussion and many previous studies of early newborn discharge, 2 important gaps in knowledge persist. First, scant information is available about how legislation mandating coverage of minimum postnatal hospital stays affects newborns' postdischarge health care use and clinical outcomes. Studies that evaluated the effect of short hospital stays before legislation was passed yielded mixed results.3,4 Some studies found short stays associated with adverse effects such as increased rehospitalizations and higher infant mortality,5–7 whereas others found that it had little or no effect on rehospitalization and other service use.8,9
Only 1 study in a health maintenance organization setting has evaluated changes in clinical outcomes associated with legislation mandating coverage of minimum postnatal hospital stays in Massachusetts.9,10 In this Massachusetts patient population, neither a policy of short hospital stays nor the legislation preventing short stays had an effect on newborn rehospitalizations, emergency department (ED) visits, or the likelihood of breastfeeding.9,10 A dearth of information exists about how the laws affected health care use and outcomes in traditional Medicaid populations, who are potentially at elevated risk of adverse outcomes after early discharge.
Second, little research has been conducted on the effectiveness of early follow-up visits, which the American Academy of Pediatrics (AAP) recommends be provided within 48 hours after discharge to all infants with short hospital stays. This recommendation is based on the clinical rationale that jaundice peaks and breast milk comes in at 72 to 96 hours after delivery. However, few studies provide direct evidence that early follow-up visits result in better health outcomes.11–13 In analyses examining outcomes after home visits11 or in special follow-up programs in 1 managed care organization,12,13 early follow-up care after a short hospital stay reduced early acute-care visits for newborns, but these studies focused on small, low-risk populations. Despite the AAP recommendations, many infants currently do not receive early follow-up visits.14–18
This study was designed to address these important information gaps in a state Medicaid population. We studied deliveries using linked Medicaid claims and birth certificate data in Ohio, where state legislation significantly reduced the proportion of mothers and newborns who experience short hospital stays.2 Our specific aims were to evaluate 1) how state legislation mandating coverage of minimum postnatal hospital stays affected receipt of follow-up visits and adverse outcomes including rehospitalization and ED visits and 2) whether early follow-up visits for newborns with short hospital stays were associated with the rates of rehospitalization, ED visits, and adverse outcomes during the first 10 days of life.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The database of Medicaid births provided information on sociodemographic characteristics of newborns and mothers, date of birth, date of discharge, postdischarge health care use, and diagnoses and procedures associated with Medicaid claims. The study cohort included normal infants who were born to Ohio Medicaid recipients who had a valid Medicaid birth claim. Normal newborns were defined as those who had diagnosis related group 391 (normal newborn), birth weight 
2000 g, and gestational age 37 weeks.
We excluded the 10% of Medicaid neonates who were born to mothers who were enrolled in health maintenance organization plans because these plans' reporting of health services use in claims data were believed to be incomplete. In Ohio, the managed care population is disproportionately nonwhite, as a result of adoption of managed care in urban areas. Although the continuous rise in managed care over the study period can affect secular trends in outcomes, it would not affect discontinuous changes studied.
The primary outcomes were rates of rehospitalization (overall and for jaundice); ED visits, including those that led to rehospitalization; and diagnoses of dehydration and neonatal infections within 10 and 21 days of birth. Rehospitalizations were based on inpatient claims, whereas ED visits, diagnoses of dehydration, and diagnoses of neonatal infections were based on outpatient claims and physician claims. On the basis of earlier work in the Ohio Medicaid population,2,8 a variable "short stay" was defined as discharge within 1 day of vaginal birth or within 2 days of cesarean birth, based on the difference between date of birth and the date of the last claim for the hospital stay. Follow-up visits within 6, 10, 14, and 21 days of birth were identified using definitions from prior work.8 In addition to home visits, follow-up visits included any visits in settings other than the ED.
We defined 3 periods related to Ohio's legislation mandating insurance coverage of minimum postnatal hospital stays: prelegislative, between July 1, 1991, and August 22, 1995 (17 quarters); legislative, between October 1, 1995, and September 30, 1996 (4 quarters); and postlegislative, between October 17, 1996, and May 30, 1998 (7 quarters). We distinguished between the legislative and postlegislative period to allow for some response to the introduction of legislation and a separate response at the implementation of the legislation. A previous study examined Ohio's legislative periods in more detail,2 and we tested the sensitivity of our analyses to alternative definitions of the time periods, but the conclusions remained unchanged. All June births were excluded because of the inability to match births and claims across fiscal years. To avoid including births from 2 different legislative periods in a single quarter, we excluded births between August 22 and September 30, 1995, and between October 1 and October 16, 1996. Quarters rather than smaller units of time were chosen as the unit of analysis to increase precision because of the relative rarity of rehospitalizations and the diagnoses of interest. Results were similar when months were used as the unit of analysis.
Statistical Analysis
Segmented linear-regression analysis19–23 was used to estimate sudden changes in levels or trends in newborns' rates of utilization and outcomes associated with the introduction and passage of legislation mandating coverage of minimum postnatal stays in Ohio. Regression models included a constant term, a term for linear time trend, and terms to estimate changes in the level or trend at the introduction and passage of legislation. In addition, models controlled for first-order autocorrelation (correlation between 2 consecutive observations). Sensitivity tests of results to alternative forms of (higher order) autocorrelation yielded quantitatively similar findings. All models were estimated for the entire cohort, as well as for the subset of infants who were delivered vaginally, and for the subset of primiparous mothers. The results using subsets of the cohort were quantitatively similar and thus are not shown.
We tested the sensitivity of our results to potential confounding of trends as a result of changes in the underlying characteristics of the mother and the neonate over time by performing the analyses described above on adjusted quarterly rates of each outcome variable. These adjusted rates come from linear models of each outcome controlling for maternal education, age, marital status, parity, race, gestational age categories, birth weight, number of prenatal care visits, metropolitan residence, and 6 previously described perinatal service/education regions24 of the state. The results from the adjusted analyses (not shown) confirm the findings using unadjusted outcomes. We computed the significance of regression coefficients using 2-tailed t tests.
For vaginally delivered newborns who were discharged within 1 day of birth, we estimated the impact of early (within 4 days of birth) follow-up visits on rehospitalization, ED visits, diagnoses of dehydration, and diagnoses of infection within 10 days of birth using instrumental variable (IV) methods.25,26 The method of IV analysis is well known in the econometric literature and increasingly used in health services research.6,27 With the use of other observational methods, estimates of the effect of early follow-up visits are vulnerable to confounding. If, for example, providers refer newborns for early follow-up visits because they are at risk of some adverse outcome not observable in administrative data, then any benefit of early follow-up visits will be underestimated. Conversely, if families who are better equipped to address the medical needs of their newborns are more likely to return with newborns for early preventive care, then the impact of early follow-up visits would be overstated.
The technique of IVs attempts to mimic the randomization that occurs in experimental settings by identifying a variable, in this case the day of week of birth, that helps to determine which newborns receive early preventive care but is not related to other potentially confounding factors. In this study, the variable of interest was a binary measure for having a follow-up visit by the age of 4 days (compared with all other vaginally delivered newborns with short stays). Visits within 48 hours of birth (by day 3 or 4, depending on the hour of birth) would coincide with the AAP's policy statement regarding care for newborns who have a short hospital stay at birth. Restricting our sample to infants who were born on weekdays, we used the day of week of birth as the instrument for receipt of an early follow-up visit by age 4 days. We excluded Saturdays and Sundays because hospital staffing patterns likely differ between weekends and weekdays and because an analysis of maternal characteristics in our sample revealed that births on weekends differed slightly from births on weekdays. For example, mothers of newborns who were born on weekends were more likely to be nonwhite and unmarried at the time of delivery.
Although we use births from all weekdays in our analyses, we explain our IV strategy using Tuesday and Thursday births as an example. Because few providers schedule early follow-up visits on a weekend, infants who are born on Tuesday and discharged on Wednesday are more likely to receive early preventive care than infants who are born on Thursday and discharged on Friday. However, infants who are born on Tuesday do not, compared with those who are born on Thursday, differ in underlying risk of adverse outcome or in the ability of their parents to attend to medical needs. Thus, differences in outcomes can be attributed to differences in early preventive care.
We performed our IV analysis in 3 steps. First, to address the concern that the timing of births by day of week might be nonrandom, we performed Pearson 
2 and analysis of variance tests of maternal characteristics at delivery (education, age, marital status, parity, race, number of prenatal care visits, region of birth, and metropolitan area residence) to confirm that no significant differences in maternal characteristics by weekday of birth existed in our sample.28 Second, we graphically summarized our outcomes as a function of the rate of early follow-up care, grouping data by day of week of the birth. Third, having excluded Saturday and Sunday births, we performed 2-stage least-squares estimation29 of our outcome variables as a function of early follow-up visits controlling for the characteristics mentioned above.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
37 weeks, and birth weight 2000 g). After excluding 11 967 June births; 2578 births between August 23, 1995, and September 30, 1995; and 939 births between October 1, 1996, and October 16, 1996, the final study cohort included 155 352 infants, including 40 225 who were vaginally delivered and had short stays.
Sample Characteristics by Legislative Period
With few exceptions, maternal demographic characteristics and medical variables changed little over the 3 legislative periods (Table 1). Mothers of newborns in the sample were slightly more educated, more likely to be primiparous, and less likely to be nonwhite in the last 2 periods than during the prelegislative period.
|
|
|
|
Figure 4 shows that rates of ED visits within 10 (21) days moved together with short stays, rising from 2.7% to 3.8% (6.1%–8.0%) during the prelegislative period (P < .001 for trends), and reversed trend when legislation was introduced. In the year after Ohio's legislation was introduced, ED use within 10 days fell from 4.2% to 3.8% (P = .088 for change in trend) and ED use within 21 days fell from 9.1% to 8.0% (P = .016 for change in trend) during the legislative period and leveled off during the postlegislative period.
|
Impact of Early Follow-up Visits on Adverse Outcomes
The analysis of the impact of early follow-up visits was conducted among infants who were delivered vaginally and had short hospital stays. A comparison of newborn characteristics found no significant differences associated with which day of the week the infant was born. As Table 2 and Fig 5 show, the likelihood of receiving an early follow-up visit varied on the basis of the day of the week the birth occurred. Only 6.5% of newborns who were delivered on Thursday received a follow-up visit within 4 days, whereas 8.9% of those who were born on Tuesdays received an early follow-up visit by age 4 days.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Comparisons With Other Studies of Early Postpartum Discharge
The current study addresses several important gaps in understanding about effective postpartum care services. Only 2 studies to date have directly evaluated the impact of length of stay legislation on outcomes using data from both the pre- and postlegislative periods, and neither focused on a large sample of newborns from a vulnerable population.9,30 The first, in Pennsylvania, found a reduction in length of postpartum stay and costs but did not evaluate subsequent outcomes.30 The second, which included 
27 000 infants in a Massachusetts health maintenance organization, found no effects of legislation on subsequent outcomes.9
The current study is unique because it evaluates health outcomes in a vulnerable population both before and after early discharge legislation. Few studies have evaluated early discharge for Medicaid-insured mothers and infants, and often these studies have had sample sizes too small to detect meaningful differences.8,31,32 Compared with the Massachusetts study,9 our research likely had better ability to detect changes in adverse outcomes after early postpartum discharge legislation because it focused on a more vulnerable, Medicaid-insured population and included >155 000 infants. Two other recent studies of early newborn discharge focused on the same Ohio Medicaid population as this study, but neither recent study linked the short-stay legislation to the outcomes evaluated here. One study evaluated outcomes during the prelegislative period, and the other analysis was limited to the relationship between short-stay legislation and the rate of short stays.2,8
The use of segmented time-series regression in this study, like the study of Madden et al,9 represents an advance over previous methods. Most existing estimates of the effect of early newborn discharge on rehospitalizations come from cross-sectional study designs, which are vulnerable to selection bias. Existing studies generally compare rehospitalization rates for infants with short hospital stays with rehospitalization rates for infants with longer stays, controlling for a limited set of demographic and medical characteristics of mothers and newborns.5,31,33–35 With 1 exception,9 studies exploiting data over time have been limited to the prelegislative period of steadily shortening hospital stays.8,36 These studies are subject to confounding when secular trends coincident with shortened hospital stays affect the outcomes of interest. For example, financial pressures could alter both length of stay at delivery and practice patterns regarding rehospitalization, masking any negative effects of short hospital stays on rehospitalization.
Findings on Newborn Early Follow-up Visits
This study provides the strongest evidence to date that early postpartum follow-up visits may reduce subsequent rehospitalizations. Potential explanations for the difference in rehospitalization related to early follow-up visits include the possibility that parents learn information in early visits that might prevent or mitigate pernicious effects of jaundice or feeding problems that can lead to readmission. An analysis of hospitalization diagnoses showed no major difference in the reason for rehospitalization between groups who had early follow-up visits and those who did not, but consistent with the literature on this population, jaundice was the most common rehospitalization diagnosis in the first 10 days of life.8
Studies of whether early postpartum outpatient visits improve infant health outcomes compared with no early postpartum follow-up visits are needed badly but are challenging to design. Several studies, including this one, suggest that many infants who have early hospital discharge do not receive follow-up visits within 48 hours, as recommended by the AAP.11–13,18,30,32 Stronger evidence about whether such visits are effective could lead physicians to schedule such visits more consistently. However, it seems unlikely that randomized trials of early follow-up visits will be undertaken because the AAP recommends that all newborns with short postpartum hospital stays receive this service.37 Among the observational study designs that are possible, the IV design in this analysis best reduces confounding. The analysis takes advantage of the variation in the likelihood of having an early follow-up visit based on the day an infant was born, which approximates a random event on Mondays through Fridays.
When early discharge legislation was introduced, some feared that the focus on increasing the hospital length of stay would detract resources from other services such as early postnatal primary care.1,38 In this Ohio Medicaid population, preventive primary care visits did not drop when legislation was introduced and passed, but continued to increase. Our study may have underestimated the percentage of infants who actually received early follow-up visits. An analysis of maternal survey data from the 2000 Pregnancy Risk Assessment Monitoring System found that 78% of Medicaid-insured mothers in Ohio reported that their infant was seen by a doctor, nurse, or other health care provider during the first week after he left the hospital (A. Lansky, W. Barfield, K. Marchi et al, unpublished data, 2004). This is higher than the 52% of infants who had visits within 14 days of birth based on our study's Medicaid claims data in 1998. The difference between these estimates is likely due to the fact that the analyses are from different years, as well as underascertainment of visits in the Medicaid data and/or overreporting of visits in the Pregnancy Risk Assessment Monitoring System survey.
Limitations
One limitation of an IV analysis of the impact of preventive care is that it necessarily focuses on a narrow group of newborns: those whose early follow-up care differs only because of the day of the week on which they were born. Although we include in our analysis all vaginally delivered newborns with a short stay regardless of when or whether they received a follow-up visit, the IV analysis effectively compares infants who received early follow-up visits by day 4 with infants receiving care in days 5 and 6, because differences in the receipt of follow-up visits that occur as a result of the day of week of birth relate primarily to differences between visits by the age of 6 days. One cannot generalize these results to care in later weeks. However, for providers and decision makers who are concerned about early rehospitalizations, preventive primary care after short stays may provide one simple and relatively inexpensive way to prevent this adverse outcome.
Our study may underestimate the rates of follow-up visits within the first 10 days of life, especially if much of the primary care in our sample was delivered through home visits that are not billed to Medicaid. However, data from the Pregnancy Risk Assessment Monitoring System suggest that approximately three-fourths of early follow-up visits in Ohio occur in office settings rather than homes (A. Lansky, W. Barfield, K. Marchi et al, unpublished data, 2004). Any underascertainment of visits due to missing data in Medicaid claims may be important for trends in follow-up visits over time but would not affect the results of our IV analysis unless it was related to the day of week of birth. If, for example, home visits not captured in the Medicaid claims were more likely to occur on weekends compared with other types of visits, the estimated rate of early follow-up visits would be understated among infants born on Thursdays and Fridays, and thus the benefit of early follow-up visits would be understated.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
We are grateful to Jeanne Madden, PhD, Dennis Ross-Degnan, ScD, and Stephen Soumerai, ScD, for helpful comments regarding the study design and analysis and for reviewing drafts of this manuscript. We thank the Ohio Department of Health Medical Technical Assistance and Policy Program for providing the linked data set.
|
FOOTNOTES |
|---|
Reprint requests to (E.M.) Department of Health Care Policy, Harvard Medical School, 180 Longwood Ave, Boston, MA 02115-5899. E-mail: meara{at}hcp.med.harvard.edu
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  C. E. Mercier, S. E. Barry, K. Paul, T. V. Delaney, J. D. Horbar, R. C. Wasserman, P. Berry, and J. S. Shaw Improving Newborn Preventive Services at the Birth Hospitalization: A Collaborative, Hospital-Based Quality-Improvement Project Pediatrics, September 1, 2007; 120(3): 481 - 488. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. M. Paul, E. B. Lehman, C. S. Hollenbeak, and M. J. Maisels Preventable Newborn Readmissions Since Passage of the Newborns' and Mothers' Health Protection Act Pediatrics, December 1, 2006; 118(6): 2349 - 2358. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Datar and N. Sood Impact of Postpartum Hospital-Stay Legislation on Newborn Length of Stay, Readmission, and Mortality in California Pediatrics, July 1, 2006; 118(1): 63 - 72. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. A. Galbraith, D. C. Grossman, T. D. Koepsell, P. J. Heagerty, and D. A. Christakis Medicaid Acceptance and Availability of Timely Follow-up for Newborns With Medicaid Pediatrics, November 1, 2005; 116(5): 1148 - 1154. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. K. Suresh and R. E. Clark Cost-Effectiveness of Strategies That Are Intended to Prevent Kernicterus in Newborn Infants Pediatrics, October 1, 2004; 114(4): 917 - 924. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Maisels and on behalf of the Subcommittee on Hyperbilirubinemi Hyperbilirubinemia Guidelines and Unintended Harms: In Reply Pediatrics, October 1, 2004; 114(4): 1135 - 1137. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2004 American Academy of Pediatrics. All rights reserved. |
||
|
||
