Impact of Medicaid managed care on utilization of obstetric care: evidence from TennCare's early years.

ABSTRACT

Background. TennCare expanded Medicaid coverage, substituting managed care for fee-for-service reimbursement in Tennessee.

Methods. To study effects of TennCare on utilization of obstetric care (office visits, prenatal tests, care at labor/delivery), we used a before (1993) and after (1995) design with North Carolina as a control state. Data came from interviews with women with various forms of insurance, delivering in 1993 or 1995 in both states. Multivariate logistic analysis was used to control for other utilization determinants (eg, demographic factors).

Results. TennCare women were only 38% as likely to have initiated prenatal care during the first trimester as those with traditional Medicaid. Ultrasonography and [alpha]-fetoprotein testing rates were higher for TennCare women, and the cesarean section rate was equivalent. However, access to care remained lower for TennCare than for the privately insured women.

Conclusions. Overall, relative to traditional Medicaid, TennCare did not adversely affect access to obstetric care during the program's early years.

**********

On January 1, 1994, Tennessee implemented TennCare, a major Medicaid enrollment expansion and placed all of its Medicaid recipients in a managed care plan--either a health maintenance organization (HMO) or preferred provider organization (PPO). (1) Although other states such as Oregon were moving in the same direction, (2) as of mid-year 1996, Tennessee was the only state to have placed all of its former fee-for-service Medicaid enrollees in full-risk, managed care plans. (3) The Arizona managed Medicaid program also has 100% of enrollees under full-risk capitation but has done so since the beginning of that program rather than diverting large numbers of enrollees who were traditionally accustomed to fee-for-service care. (4) Tennessee's decision was all the more remarkable given the state's relative lack of previous Medicaid managed care experience. (4,5) Yet, at the same time, Tennessee has achieved much more success than other states in reducing the numbers of lower income adults without insurance coverage . (6)

Three important benefits are potentially derived from this state experiment. First, capitation fixes government cost, given the number of recipients. TennCare's capitation rates are budget based rather than market based. The

state contracts with managed care organizations (MCOs) to provide medically necessary services to program enrollees. The MCOs, which during TennCare's history have included a large PPO and several HMOs, are responsible for developing a network of health care providers and negotiating payment rates with individual providers. To date, rates have been set well bellow per capita expenditures previously incurred by Medicaid, despite the substantial discounts Medicaid obtained from providers. The state has the authority, which it has exercised, to limit the number of enrollees. Second, by placing plans at risk of incurring overruns, TennCare creates incentives for plans and providers to control costs. Third, TennCare allows greater access to coverage by allowing the uninsured to buy into the pr ogram. In addition to persons satisfying Medicaid eligibility requirements before program implementation, TennCare has covered persons uninsurable because of a previous health condition and those not eligible for coverage through an employer-sponsored or government-sponsored plan as of a specified date. During the past decade, coverage of the uninsured was much less important for obstetric than other forms of care because of previous Medicaid expansions for pregnant women and infants.

As long as Tennessee can achieve its expenditure targets and MCOs continue to participate, an important question concerns cost containment. In this article, we assess effects of the TennCare experiment on utilization of obstetric services during its early years, 1994 to 1997. Nationally, Medicaid covered 39% of births in 1994. (7) Thus, understanding the effects of Medicaid managed care on pregnant women is critical to an overall evaluation of the program. Obstetric care provides a "window on the system." Unlike other types of care, many important effects of care are plausibly more immediate. There is no lengthy latency period, as for example, with cancer or heart disease, where failure to diagnose and treat may not be manifested in outcome data for several years.

This study addresses two basic issues. First, how did utilization of office visits, prenatal tests, and patterns of care at labor and delivery compare between TennCare and traditional Medicaid during the program's early years? Second, compared with persons having other forms of coverage or no coverage, did TennCare enrollees gain access to "mainstream" obstetric care? We contrasted utilization by TennCare enrollees with that of patients who had private coverage and the minority of patients who had no insurance.

STUDY AND METHODS

Control State

In late 1996 and early 1997, we surveyed community hospitals in Tennessee and North Carolina. Although our analysis focused on Tennessee, we included North Carolina as a control group to better isolate changes attributable to TennCare. Most importantly, the states used the same eligibility standards for pregnant women and infants in the two states. However, North Carolina used virtually no Medicaid managed care at the time, whereas in Tennessee, Medicaid managed care rose from a negligible amount to 100% as a result of TennCare implementation. Thus, the control group methodology captured the effects of the altered financial incentives created by TennCare.

In general, hospitals in the two states were similar in terms of bed size and teaching status. North Carolina hospitals in 1993 were slightly larger and had slightly more residents per bed. The only notable difference was in ownership composition, with Tennessee having a much larger share of for-profit hospitals. We accounted for this difference by excluding for-profit hospitals from our study.

According to birth certificate data from both states in 1993, the states were similar in terms of maternal age, education, and marital status. North Carolina had a higher percentage of blacks and other nonwhites. Per capita income in the two states was virtually identical, with Tennessee ranking 36th and North Carolina ranking 34th among all states in 1996. (8) Medicaid covered 47% of births in Tennessee and 45% of births in North Carolina in 1993; in 1995, the corresponding percentages were 48% and 44%. (9)

DATA COLLECTION

Hospital Sample

Hospitals were selected so that each Tennessee hospital would have a North Carolina counterpart. In Tennessee, 15 hospitals were contacted, and two declined to participate. In North Carolina, 14 hospitals were contacted, and 4 declined to participate (Fig 1). The hospitals included in this study in the two states were almost identical in terms of size and in the ratio of residents to beds (0.084 in Tennessee, 0.10 in North Carolina). In terms of Medicaid inpatient days to total hospital beds, Tennessee was slightly higher than North Carolina (9.1 in Tennessee versus 8.3 for North Carolina). Hospitals were located in all parts of each state and were in metropolitan and nonmetropolitan locations.

Patient Samples

After receiving approval from our Institutional Review Board, we asked each hospital for names, addresses, and telephone numbers for patients discharged during 1993 and 1995 with normal births (DRG 391) and high-risk births (DRG 385-90), with preference given to discharges in the fourth quarters of both years. We used North Carolina hospital discharge data for 1993 to specify samples for each hospital by year, diagnosis, and payer status to allow us to develop desired sample sizes for these groups. Reliable discharge data were not available for Tennessee. We matched Tennessee hospitals with comparable hospitals in North Carolina and doubled the samples requested.

With lists provided by the hospitals, staff contacted individuals to obtain consent for interviews. From the 4,209 persons for whom contacts were sought, 986 mothers were interviewed.

The majority of the 986 respondents had private insurance at the time of delivery (Fig 2). A substantial fraction of the 1995 sample was enrolled in TennCare (37.5%). None were enrolled in this program in 1993, since this was the year immediately preceding its implementation. Traditional Medicaid includes Medicaid enrollees from Tennessee in 1993 and from North Carolina in 1993 and 1995.

The low number of uninsured mothers (around 3% of the sample) reflects Medicaid expansions in both states already in place by 1993. There are several reasons for the difference between the number of initial contacts and final interviews. Six of the 23 participating hospitals required that hospital staff obtain permission directly from patients. This additional step reduced the participation rate of patients from these hospitals. For these hospitals, we were entirely reliant on hospital staff effort to obtain permission to participate; by contrast, for the remaining hospitals, obtaining consent was under our direct control. When the hospital was an intermediary in obtaining permission to participate, we could not determine the reason for nonresponse. For patients in the remaining hospitals, 3.6% of Tennesseans and 9.3% of North Carolinians refused to be interviewed. Otherwise, the patient could not be reached by telephone because the number was unlisted or there was no answer, even after at least eight attempt s. Medicaid and self-pay patients had lower refusal rates than those with private insurance or Medicare. The organization that conducted the survey estimated that 10% of persons sampled had no telephone. A plausible inference is that most of the patients we were unable to locate had moved. Persons in this age cohort tend to be highly mobile.

The survey format asked respondents to provide information about both their labor/delivery experience (the index admission) and their current situation. Information was obtained on utilization of services before delivery, selected prenatal tests, and utilization at or after delivery.

Patient demographic characteristics from our sample were similar to those of patients from birth records in the two states in 1993 and 1995 in some respects, but there were differences in others (Table 1). The sample was almost identical to the population in terms of percent married and years of schooling. Since we intentionally oversampled high-risk births, the largest difference was in terms of the age distribution and child characteristics. Our sample, by design, contained appreciably higher proportions of mothers 35 and over, and higher proportions of low and very low birth weight infants.

ANALYSIS

Dependent Variables

We evaluated three groups of dependent variables. First, five binary variables were used to identify utilization of services before delivery: had prenatal care initiated in the first trimester; had a regular provider of prenatal care; had usual private source of care in an office; had a nurse as the usual prenatal care provider; and was referred to another doctor during pregnancy. The second group of dependent variables were prenatal tests: ultrasound, amniocentesis, glucose tolerance test, and [alpha]-fetoprotein test. The third group of dependent variables included measures to quantify utilization at and after delivery: cesarean section, delivery by a regular doctor or affiliate, hospital stay of 1 day or less, and subsequent rehospitalization of the infant.

Effect of Payer Status

As a first line of controls, we included a binary variable for 1993 to control for time-related changes in utilization, and a binary for Tennessee to control for time-invariant state effects. Other payer status variables were for private insurance and no insurance. Traditional Medicaid was the omitted reference group.

The effect of TennCare on utilization was measured relative to traditional Medicaid as it existed in Tennessee in 1993 and in North Carolina in 1993 and 1995. The TennCare effect was obtained directly from the coefficient on a variable for TennCare payer status. Other forms of insurance used in the analysis also included traditional Medicaid (the omitted reference group), uninsured, or other insurance (private or Medicare).

Other Covariates

The other factors that we controlled for in our analysis (all defined at the index admission) are the following: Demographics included maternal age (<20, 21 to 34, or [greater than or equal to]35); black, white, or Hispanic race; years of schooling; marital status; and total number of persons in household. Financial variables included current family income and family income missing. The mother's health before delivery was self-rated as good, fair, or poor during the year before becoming pregnant, relative to excellent or very good health. Child-specific health was defined by parity and a binary variable denoting when a physician told the mother to expect problems with the child after delivery. Finally, variables describing problems during delivery included inadequate oxygen, infection, umbilical cord around neck or knotted, excessive bleeding, child too large, placenta previa, and other problems. When appropriate to the dependent variable, statistical analysis included additional information on child-specific health--multiple gestation, birth more than 4 weeks overdue, low birth weight (1,500 to 2,500 g), and very low birth weight (<1,500 g). A "family income missing" variable identified respondents for whom no income information was provided. In such cases, current income was set to zero.

Estimation 

All dependent variables were binary. Therefore, we used logistic regression to estimate all equations.

RESULTS

Utilization Before Index Admission: Prenatal Care

In two respects, pregnant women enrolled in TennCare in 1995, the second full year of the program, obtained less care than did those enrolled in traditional Medicaid (Table 2). Compared with mothers on traditional Medicaid, TennCare enrollees were 38% as likely to have initiated prenatal care during the first trimester (P = .05) than their counterparts enrolled in a traditional Medicaid program.

The uninsured had lower rates of use than those on TennCare or with private insurance on all measures. Compared with the omitted reference group, recipients on traditional Medicaid, the uninsured were 25% as likely to start prenatal care in the first trimester and 16% as likely to be referred to another doctor during pregnancy. These results were not statistically significant at conventional levels.

Compared with traditional Medicaid, privately insured women had much better access to prenatal care. Privately insured women were more than six times more likely to have a source of private care in an office (P < .001) and only half as likely to have had a usual provider who was a nurse (P = .01).

Most of the covariates for demographic characteristics of the mother, family income, and maternal health and pregnancy history had plausible effects on the dependent variables. Never married respondents were less likely to initiate care in the first trimester (P < .001). Black or Latin American mothers were less than half as likely to have had a prenatal provider than whites (P = .02). Higher income mothers were more likely to have an office-based private source of care (P = .01). Women whose delivery was 2 weeks late or more were more likely to have a nurse as the usual provider of prenatal care, but at the same time, they were more likely to have had a referral to another physician during the pregnancy (P < .001). The year 1993 odds ratios of 0.38 for initiation of care in the first trimester (P = .01) and 0.60 for referral to another physician during pregnancy (P = .02) may indicate recall bias about these dimensions of care.

We did a complete "difference-indifference-in-difference" specification that included all the standard second-order interactions between year, state, and coverage dummies. (10) This altered few of the TennCare results, but produced implausible upper bounds for many of the confidence intervals on the interaction terms--a result attributable to small cell problems created by the introduction of these interactions (results available from authors on request). Therefore, we show only the simplified specification.

Prenatal Tests

In general, prenatal testing was more extensive under TennCare than under traditional Medicaid. The differences were particularly noteworthy for ultrasonography and [alpha]-fetoprotein testing (Table 3). Ultrasonography was more than twice as likely for TennCare than for traditional Medicaid recipients (P = .04), and [alpha]-fetoprotein testing was almost twice as likely for TennCare enrollees (P = .05). Among the covariates, none of the odds ratios for the 1993 variable were statistically significant at conventional levels. The odds ratio for glucose tolerance testing was 1.40, suggesting much higher use for TennCare than for traditional Medicaid. However, the result was not statistically significant at conventional levels. For amniocentesis, the odds ratio implies lower use under TennCare, but this result was also not statistically significant.

Utilization At and After Labor/Delivery

Overall, utilization patterns at labor and delivery of TennCare enrollees and those on traditional Medicaid were statistically indistinguishable (Table 4). TennCare mothers were 15% more likely to have stayed in the hospital for 1 day or less, but the difference was not statistically significant. Privately insured patients were only 95% as likely to have stayed for a day or less than those with traditional Medicaid, but this difference was not statistically significant either. In terms of patterns of care at labor/delivery, the care of the uninsured did not differ from mothers enrolled in traditional Medicaid or in TennCare.

We found no difference between TennCare and traditional Medicaid in the probability of a cesarean section, having a regular doctor or affiliate deliver the child, staying 1 day or less in the hospital, or having a child who was readmitted to the hospital (after initial discharge). None of the four odds ratios for TennCare status was substantially below 1.00.

The only measure on which all pregnant women without private health insurance were at a clear disadvantage relative to those with private insurance was in having a regular doctor (or affiliate of the regular doctor) deliver the child. For this measure, the odds ratio on private insurance was 3.56 (P< .001).

The odds ratio from the analysis of a probability of staying in the hospital for 1 day or less was 0.59 for the 1993 binary variable (P = .01) suggesting substantial reductions in maternal length of stay after 1993. However, these reductions apparently were across the board, not limited to TennCare or any other payer group.

As with Table 2, we performed a complete "difference-in-difference-in-difference" specification that altered few of the TennCare results for utilization at and after labor and delivery. For reasons mentioned, we again show only the simplified specification results (available from authors on request).

DISCUSSION

We found that in the second year after implementation, TennCare reduced some types of utilization before labor and delivery; in particular, it delayed initiation of care but did not decrease prenatal testing. In fact, TennCare recipients were more likely than those covered by traditional Medicaid to receive several types of tests. We found no effects of TennCare on utilization at labor and delivery or after discharge from the hospital. In some respects, for prenatal testing, TennCare did provide a key to the door of mainstream medicine. In other respects, however, a two-tier delivery system remained. In particular, TennCare recipients were far less likely than privately insured women to have an office-based provider.

Initiation of care in the first trimester was particularly problematic for TennCare enrollees relative to those with traditional Medicaid coverage. Late initiation of care might have been seen as a temporary problem of TennCare implementation; most of the women on TennCare in our sample became pregnant in late 1994 or 1995, less than a year on average after the program was implemented (January 1, 1994). In another study using data from birth records from Tennessee and North Carolina for the same period, (11) we found delays in initiation of care in Tennessee relative to North Carolina in 1995. However, in absolute terms, the percentage of pregnant Tennessee women initiating care during the first trimester increased, but not as quickly as first-trimester care initiation in North Carolina. We attribute this difference to an access barrier under TennCare, since eligibility criteria in the two states were perfectly matched.

* Our results differ from recent research on the effects of managed care on labor and delivery. Late initiation of prenatal care of Medicaid enrollees relative to others was reported in a recent study of patients in Michigan. (12) In our study, being covered by traditional Medicaid made no difference in this respect. A recent study of TennCare based on vital statistics found no change in the percentage of pregnant women on Medicaid or TennCare who initiated prenatal care after the fourth month of pregnancy. (13) That study used no control group to quantify TennCare's effects, making it difficult to separate changes from TennCare implementation with general shifts in medical practices.

Many "nonfindings" about TennCare are also noteworthy. Among these are our results for frequency of 1-day stays and cesarean sections. The increased frequency of 1-day maternity stays among managed care patients has attracted considerable attention in the media and among policy makers. In this regard, our finding of no statistically significant difference between TennCare and traditional Medicaid is interesting. In this sense, the accusations that Medicaid managed care, or managed care more generally, has cut corners are inconsistent with this empirical evidence. Our results for cesarean sections are not consistent with those of a recent review of studies showing that women in HMOs generally have lower rates of cesarean section than those with private coverage. (14) A more recent study of cesarean section rates using California data not included in this review showed that HMO versus non-HMO enrollment had no effect on cesarean section rates. (15) The difference may reflect relative growth of HMO types in the 1990s that are less prone to restrict utilization.

The access-enhancing record of TennCare is a case of "half empty or half full," depending on one's perspective. Compared with being uninsured, TennCare provided some gains in terms of having access to a regular prenatal care provider. However, after Medicaid expansions of the early 1990s, relatively few pregnant women were uninsured. In some cases, the improved access was to a nurse rather than a physician. Prenatal visit frequency was much greater for TennCare recipients than for the uninsured. Of course, although we controlled for income, education, health status, and other demographic characteristics, we did not measure patients' preferences for health care. Lower rates of use among the uninsured are possibly attributable to a perceived lack of efficacy of such services rather than to the TennCare program perse.

From another perspective, however, access in some respects was not nearly as good for TennCare enrollees as it was for the privately insured. In particular, access to a private source of prenatal care was not nearly as good for TennCare mothers as for privately insured mothers. Initiation of prenatal care by TennCare enrollees was often much later than among even traditional Medicaid recipients.

We found no empirical evidence to support the view that Medicaid managed care restricted access to specialized prenatal care services. If anything, providers supplied more of the same. Our finding of no reduction in specialized tests is consistent with evidence from a comparison of HMO-fee-for-service obstetric treatment patterns in King County Washington. That study was limited to privately insured patients. (16)

A critical policy concern is whether utilization reductions either translate into cost savings, in which case they might be viewed favorably, or instead are achieved at the expense of good health, in which case the interpretation is considerably different. Recent evidence from Kaiser Permanente showed that prenatal visit rates can be reduced by as much as 20% without adversely affecting birth outcomes. (17)

An even more important policy issue is whether TennCare infants have worse health outcomes. The evidence on the whole is favorable to TennCare, at least for the period immediately after birth. At a minimum, TennCare did not harm birth outcomes. A TennCare study based on vital records also found that perinatal outcomes for Medicaid eligibles did not change between 1993 and 1995. (15) In another study, we assessed the effects of TennCare on birth outcomes for the same observational period used in this study. (11) Data came from birth records merged with death records from Tennessee and North Carolina for both 1993 and 1995. This file contained more than 300,000 records. Unfortunately, payer states were not identified on the records. Thus, any conclusions regarding the effect of TennCare were necessarily inferences from aggregate data that did not distinguish TennCare from non-TennCare births.

In that study, we found no statistically significant differences in infant mortality when Tennessee in 1995 was compared with North Carolina or when Tennessee in 1995 was compared with Tennessee in 1993 (pre-TennCare). This difference remained when we looked at a geographically based "poverty" subsample. Furthermore, we found no statistically significant differences in the following areas of risk: (1) risk of an abnormal condition at birth; (2) risk of low or very low birth weight; and (3) risk of transfer of the infant to a facility other than the facility at which the delivery occurred.

Evidence on the influence of Medicaid managed care on birth outcomes from other states is, if anything, encouraging. A study of 1994 births in Wisconsin found no difference in birth outcomes for Medicaid patients enrolled in managed care programs compared with others who remained in traditional fee-for-service care, and patients in managed care were more likely to have received adequate prenatal care. (18) Among Medi-Cal women in California, the probability of low birthweight was lower in a capitated payment than in a fee-for-service comparison group, holding several other factors constant. (19)

Finally, several caveats are in order. First, we studied the program only during its first 3 years. Since then, more than half a decade has elapsed. Public programs are living entities and are subject to important changes.

By evaluating the program too early, utilization control may not have been fully evident because utilization management programs may not have been fully implemented. Also, improvement in access may not be fully evident because of the initial difficulties in enrollment procedures due to the rapid startup. Although these are possibilities, the earliest index admissions for TennCare patients in our study were more than a year after TennCare was implemented. On the other hand, as time passes, the effect of an intervention such as TennCare becomes muddled by other influences.

The second issue relates to potential recall bias. Given that our surveys required recall of events that in some cases were several years earlier, can our results be explained by recall bias? Although we cannot rule out recall bias, such bias should be limited for four reasons. First, the usage rates for several self-reported variables are consistent with data from birth records. In particular, we were able to cross check the recall of prenatal tests, which should be most affected by recall bias. The cesarean section rates and rates of use of ultrasonography and amniocentesis are consistent with birth record data. Second, a previous study of maternal recall of births 5 to 7 years earlier showed 89% agreement between women's responses and their charts. (20) Third, we included a covariate for 1993 to account for recall bias. The results suggested possible recall bias in some dimensions of prenatal care, but the covariate for 1993 should have accounted for this. Fourth, the most important recall issue for this a nalysis is differential rates of recall by payer. Thus, the key issue is not whether women may have overestimated or underestimated their rates of use in general, but whether there was differential recall between TennCare and traditional Medicaid respondents in particular.

A third issue is possible nonrepresentative responses. Here again, the most important issue is whether we obtained representative samples within payer groups. If, for example, we obtained relatively well-off TennCare recipients, we may not have found some problems that truly existed. We controlled for income, race, and education, making this less likely, but we cannot know with certainty that we have completely eliminated bias from this source.

A fourth issue is statistical power. Our total sample of 986 births is small in comparison with data bases from birth or claims records. With a larger sample, we may have found more statistically significant differences in general utilization before labor/delivery that might have been disadvantageous to TennCare relative to traditional Medicaid. A further issue concerns the traditional Medicaid comparison sample, which was slightly overweighted toward Tennessee in 1993.

Fifth, our results for TennCare reflect average effects. The MCOs that have contracted with TennCare have been heterogeneous. Some appear to have performed adequately, others less so. To gauge performance of individual MCOs would have required a much larger sample than the one we collected.

The final issue is applicability of these findings. Generalizing the experiences of a single state may be invalid. In a sense, TennCare is unique, particularly in the rapidity of implementation and in terms of the expansion of public coverage, but the latter feature should generally have made it a more effective instrument of utilization control.

Many of the findings are consistent with those of previous research. Tennessee is not unique in having substantial excess capacity and the sizable purchasing advantage associated with a program as large as Medicaid. The relative immaturity of the managed care market in Tennessee may have made plans more willing to accept low capitation payments to quickly gain covered lives. (1) If so, such discounts would have created stronger incentives to curtail use than might occur in other states with more mature markets.

TennCare represents an interesting experiment by a state to address the rapidly rising outlays for Medicaid it had experienced. Subsequent to its enactment, the state achieved major reductions in the number of persons without health insurance. For this reason, many states will be monitoring with great interest the program's experience as it matures. Our results suggest that observers concerned with prenatal care may not need to be concerned about utilization changes at the time of labor/delivery or about the adequacy of prenatal testing. However, managed Medicaid programs that follow the TennCare example may require additional support to open the door to mainstream medical care for managed Medicaid enrollees. Continued attention to TennCare utilization will provide further lessons in managed Medicaid policy, for both Tennessee and the remainder of the country.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

TABLE 1

Charactertistics of Our Sample Versus Characteristics of All Births

                                              Sample             Birth
                                                                Records
                                            Population         (Tenn and
                                                                  NC)
Variable                                 Mean       SD        Mean

Demographic characteristics

  Age
     Under 20                             3.5      18.3         17.0
     21 to 34 *                          70.2      45.8         73.9
     [greater than or equal to]35        25.4      43.5          9.1
  Ethnicity (%)
     Black                               19.9      39.9         25.7
     White *                             76.6      42.3         69.8
     Hispanic                             1.8      13.5          2.3
  Marital status (%)
     Married *                           69.1      46.3         68.0
     Divorced/separated/widowed           5.0      21.9          -
     Never married                       25.7      43.7          -
     Years of schooling                  13.7       2.2         12.9
  Household characteristics
     Other people in household            2.9       1.5          -
     Years lived in area                 10.9      10.3          -
  Ecnomic characteristics
     Family income (thousands)           38.7      33.3          -
     Employed at index admission(%)      70.0      45.9          -
  Year and state
     1993                                42.8      49.5         49.6
     Tennessee                           75.0      43.3         42.3
  Baseline health status
     Excellent or very good              72.1      44.9          -
      health *
     Good, fair or poor health           27.8      44.8          -
  Birth variables
     Told to except problem with         20.9      40.7          -
      delivery
     First child                         45.9      49.9         35.7
     Multiple gestation pregnancy         3.2      17.6          2.7
     Late delivery                        3.8      19.1          6.4
     Normal (not high risk) birth        56.6      49.6          -
  Child characteristics
     Low birth weight (<2,500 g)         13.0      33.6          7.1
     Very low birth weight                4.7      21.2          2.0
     (<1,500 grams)
     No. *                                976                318,235

                                       Birth
                                      Records
                                     (Tenn and
                                        NC)
Variable                               SD

Demographic characteristics
  Age
     Under 20                         37.6
     21 to 34 *                       43.9
     [greater than or equal to]35     28.8
  Ethnicity (%)
     Black                            43.7
     White *                          45.9
     Hispanic                         15.0
  Marital status (%)
     Married *                        46.6
     Divorced/separated/widowed         -
     Never married                      -
     Years of schooling                4.5
  Household characteristics
     Other people in household          -
     Years lived in area                -
  Ecnomic characteristics
     Family income (thousands)          -
     Employed at index admission(%)     -
  Year and state
     1993                             50.0
     Tennessee                        49.4
  Baseline health status
     Excellent or very good             -
      health *
     Good, fair or poor health          -
  Birth variables
     Told to except problem with        -
      delivery
     First child                      47.9
     Multiple gestation pregnancy     16.3
     Late delivery                    24.4
     Normal (not high risk) birth       -
  Child characteristics
     Low birth weight (<2,500 g)      25.6
     Very low birth weight            14.1
     (<1,500 grams)
     No. *

* Omitted reference group for regression analysis.

Dashes indicate that variable in question was not applicable to this
sample group.
TABLE 2

General Utilization Before Index Hospitalization

                                            Initiated Care in
           Explanatory                       First Trimester
            Variables                       OR            95% CI

Health insurance
  TennCare                                 0.38         (0.15-0.98)
  Uninsured                                0.25         (0.06-1.03)
  Private                                  0.90         (0.42-1.93)

Demographics
  Age <20                                  0.47         (0.18-1.24)
  Aged [greater than or equal to]35        0.58         (0.31-1.10)
  Black or Hispanic                        0.89         (0.50-1.59)
  Divorced/separated/widowed               0.36         (0.14-0.93)
  Never married                            0.31         (0.16-0.61)
  Years of education                       1.12         (0.98-1.28)
  Total people in household                0.90         (0.80-1.02)

Income
  Current income                           1.04         (0.92-1.17)
  Income data missing                      1.07         (0.42-2.71)

Health and pregnancy history
  First pregnancy                          1.17         (0.68-2.00)
  Told to expect birth problem             1.05         (0.56-1.94)
  Multiple gestation                        --              --
  Two weeks late or more                    --              --
  In good, fair, or poor health            0.62         (0.38-1.03)

Other
  Year 1993                                0.38         (0.18-0.81)
  Tennessee                                0.88         (0.45-1.73)
Percent "yes"                                              91.4

                                     Initiated      Had Regular
                                      Care in        Prenatal
           Explanatory                 First       Care Provider
                                     Trimester
            Variables                 P              OR

Health insurance
  TennCare                           .05            0.64
  Uninsured                          .06            0.29
  Private                            .79            2.72

Demographics
  Age <20                            .13            1.08
  Aged [greater than or equal to]35  .09            1.01
  Black or Hispanic                  .68            0.40
  Divorced/separated/widowed         .03            0.60
  Never married                      .00            1.24
  Years of education                 .11            1.13
  Total people in household          .11            0.97

Income
  Current income                     .52            1.07
  Income data missing                .89            0.58

Health and pregnancy history
  First pregnancy                    .58            1.14
  Told to expect birth problem       .89            0.78
  Multiple gestation                 --              --
  Two weeks late or more             --              --
  In good, fair, or poor health      .07            0.90

Other
  Year 1993                          .01            0.94
  Tennessee                          .71            2.21
Percent "yes"

                                      Had Regular Prenatal     Had Usual
                                                                Private
           Explanatory                    Care Provider        Source in
                                                                Office
            Variables                  95% CI      P           OR

Health insurance
  TennCare                           (0.19-2.15)  .47          0.92
  Uninsured                          (0.07-1.31)  .11          0.36
  Private                            (0.93-7.97)  .07          6.15

Demographics
  Age <20                            (0.26-4.62)  .91          0.70
  Aged [greater than or equal to]35  (0.39-2.60)  .98          1.59
  Black or Hispanic                  (0.19-0.88)  .02          0.68
  Divorced/separated/widowed         (0.18-2.02)  .41          0.80
  Never married                      (0.51-3.02)  .64          0.55
  Years of education                 (0.93-1.38)  .21          1.09
  Total people in household          (0.76-1.24)  .81          1.05

Income
  Current income                     (0.88-1.30)  .49          1.36
  Income data missing                (0.20-1.66)  .31          1.19

Health and pregnancy history
  First pregnancy                    (0.54-2.41)  .73          1.53
  Told to expect birth problem       (0.34-1.78)  .56          1.51
  Multiple gestation                     --       --            --
  Two weeks late or more                 --       --            --
  In good, fair, or poor health      (0.45-1.80)  .76          1.41

Other
  Year 1993                          (036-2.42)   .89          0.51
  Tennessee                          (0.91-5.35)  .08          3.60
Percent "yes"                           95.8

                                        Had Usual Private
           Explanatory                  Source in Office
            Variables                  95% CI      P

Health insurance
  TennCare                           (0.35-2.40)  .86
  Uninsured                          (0.10-1.23)  .10
  Private                            (2.67-14.2)  .00

Demographics
  Age <20                            (0.25-1.91)  .48
  Aged [greater than or equal to]35  (0.69-3.67)  .28
  Black or Hispanic                  (0.39-1.18)  .17
  Divorced/separated/widowed         (0.28-2.30)  .68
  Never married                      (0.29-1.04)  .07
  Years of education                 (0.94-1.27)  .27
  Total people in household          (0.86-1.28)  .66

Income
  Current income                     (1.10-1.69)  .01
  Income data missing                (0.49-2.87)  .70

Health and pregnancy history
  First pregnancy                    (0.88-2.65)  .13
  Told to expect birth problem       (0.74-3.10)  .26
  Multiple gestation                     --       --
  Two weeks late or more                 --       --
  In good, fair, or poor health      (0.84-2.39)  .20

Other
  Year 1993                          (0.23-1.12)  .09
  Tennessee                          (1.82-7.11)  .00
Percent "yes"                           90.3

                                             Usual Provider
           Explanatory                         Was a Nurse
            Variables                       OR           95% CI

Health insurance
  TennCare                                 1.34        (0.75-2.39)
  Uninsured                                0.35        (0.09-1.28)
  Private                                  0.52        (0.32-0.86)

Demographics
  Age <20                                  1.37        (0.60-3.11)
  Aged [greater than or equal to]35        1.10        (0.74-1.66)
  Black or Hispanic                        0.96        (0.64-1.44)
  Divorced/separated/widowed               0.87        (0.42-1.81)
  Never married                            1.27        (0.81-1.81)
  Years of education                       1.00        (0.91-1.09)
  Total people in household                0.98        (0.87-1.11)

Income
  Current income                           0.99        (0.93-1.06)
  Income data missing                      1.23        (0.67-2.27)

Health and pregnancy history
  First pregnancy                          0.84        (0.59-1.19)
  Told to expect birth problem             0.86        (0.58-1.29)
  Multiple gestation                       0.74        (0.27-2.02)
  Two weeks late or more                   3.98        (1.96-8.07)
  In good, fair, or poor health            0.84        (0.59-1.20)

Other
  Year 1993                                1.09        (0.73-1.63)
  Tennessee                                0.62        (0.42-0.91)
Percent "yes"                                             22.6

                                       Usual        Referred to
                                      Provider        Another
           Explanatory                 Was a       Doctor During
                                       Nurse         Pregnancy
            Variables                 P              OR

Health insurance
  TennCare                           .32            0.54
  Uninsured                          .11            0.16
  Private                            .01            0.85

Demographics
  Age <20                            .45            0.71
  Aged [greater than or equal to]35  .63            1.71
  Black or Hispanic                  .84            0.88
  Divorced/separated/widowed         .71            1.13
  Never married                      .30            0.85
  Years of education                 .92            1.07
  Total people in household          .79            0.98

Income
  Current income                     .79            1.02
  Income data missing                .51            0.98

Health and pregnancy history
  First pregnancy                    .32            1.17
  Told to expect birth problem       .47            3.75
  Multiple gestation                 .56            0.98
  Two weeks late or more             .00            3.17
  In good, fair, or poor health      .34            1.50

Other
  Year 1993                          .68            0.60
  Tennessee                          .01            1.20
Percent "yes"

                                       Referred to Another
           Explanatory               Doctor During Pregnancy
            Variables                  95% CI      P

Health insurance
  TennCare                           (0.27-1.08)  .08
  Uninsured                          (0.02-1.31)  .09
  Private                            (0.47-1.51)  .57

Demographics
  Age <20                            (0.22-2.35)  .58
  Aged [greater than or equal to]35  (1.14-2.55)  .01
  Black or Hispanic                  (0.54-1.45)  .62
  Divorced/separated/widowed         (0.51-2.52)  .76
  Never married                      (0.48-1.48)  .56
  Years of education                 (0.97-1.17)  .20
  Total people in household          (0.85-1.12)  .76

Income
  Current income                     (0.96-1.09)  .47
  Income data missing                (0.46-2.08)  .95

Health and pregnancy history
  First pregnancy                    (0.79-1.73)  .43
  Told to expect birth problem       (2.61-5.40)  .00
  Multiple gestation                 (0.38-2.51)  .96
  Two weeks late or more             (1.47-6.84)  .00
  In good, fair, or poor health      (1.02-2.19)  .04

Other
  Year 1993                          (0.40-0.91)  .02
  Tennessee                          (0.79-1.82)  .40
Percent "yes"                           19.2

Statistical analysis used the "proc logistic" command of SAS 6.12 with
the risk limit (/rl) option.

OR = Odds ratio, CI - confidence interval.
TABLE 3

Prenatal Tests


            Explanatory                           Ultrasound
             Variables                    OR         95% CI      P

Health Insurance
  Tenn Care                              2.22      (1.03-4.77)  .04
  Uninsured                              0.72      (0.21-2.46)  .60
  Private                                1.59      (0.86-2.95)  .14

Demographics
  Age <20                                1.67      (0.56-5.05)  .36
  Aged [greater than or equal to]35      1.05      (0.60-1.82)  .87
  Black or Hispanic                      0.51      (0.32-0.83)  .01
  Divorced/separated/widowed             0.52      (0.22-1.26)  .15
  Never married                          0.54      (0.30-0.96)  .04
  Years of education                     1.07      (0.96-1.21)  .23
  Total people in household              0.87      (0.77-0.98)  .03

Income
  Current income                         1.00      (0.92-1.09)  .99
  Income data missing                    0.47      (0.23-0.98)  .04

Health and pregnancy history
  First pregnancy                        1.03      (0.66-1.62)  .90
  Told to expect birth problem           1.28      (0.73-2.22)  .39
  Multiple gestation                      --           --       --
  Two weeks late or more                  --           --       --
  In good, fair, or poor health          1.56      (0.96-2.54)  .07

Other
  Year 1993                              0.86      (0.51-1.47)  .58
  Tennessee                              0.60      (0.35-1.04)  .07

Percent "yes"                                         88.0

                                             Glucose Tolerance
            Explanatory                             Test
             Variables                       OR             95% CI

Health Insurance
  Tenn Care                                 1.40          (0.78-2.52)
  Uninsured                                 0.55          (0.20-1.51)
  Private                                   1.04          (0.64-1.71)

Demographics
  Age <20                                   1.36          (0.58-3.19)
  Aged [greater than or equal to]35         0.97          (0.64-1.45)
  Black or Hispanic                         0.44          (0.30-0.64)
  Divorced/separated/widowed                0.85          (0.41-1.74)
  Never married                             0.79          (0.50-1.23)
  Years of education                        1.10          (1.01-1.21)
  Total people in household                 0.93          (0.84-1.02)

Income
  Current income                            1.03          (0.96-1.11)
  Income data missing                       0.37          (0.21-0.66)

Health and pregnancy history
  First pregnancy                           1.04          (0.74-1.47)
  Told to expect birth problem              1.29          (0.86-1.93)
  Multiple gestation                         --               --
  Two weeks late or more                     --               --
  In good, fair, or poor health             1.06          (0.74-1.50)

Other
  Year 1993                                 0.86          (0.58-1.28)
  Tennessee                                 0.50          (0.33-0.77)

Percent "yes"                                                76.2

                                      Glucose      Alpha Fetoprotein
                                     Tolerance
            Explanatory                 Test             Test
             Variables                P                OR

Health Insurance
  Tenn Care                          .26              1.74
  Uninsured                          .24              0.62
  Private                            .86              1.73

Demographics
  Age <20                            .48              1.63
  Aged [greater than or equal to]35  .87              1.27
  Black or Hispanic                  .00              0.55
  Divorced/separated/widowed         .65              0.70
  Never married                      .29              0.70
  Years of education                 .03              1.14
  Total people in household          .11              0.92

Income
  Current income                     .36              1.03
  Income data missing                .00              0.65

Health and pregnancy history
  First pregnancy                    .83              1.12
  Told to expect birth problem       .23              1.48
  Multiple gestation                 --                --
  Two weeks late or more             --                --
  In good, fair, or poor health      .76              1.20

Other
  Year 1993                          .45              0.80
  Tennessee                          .00              0.78

Percent "yes"

                                        Alpha Fetoprotein
            Explanatory                       Test
             Variables                 95% CI      P

Health Insurance
  Tenn Care                          (1.00-3.04)  .05
  Uninsured                          (0.21-1.87)  .40
  Private                            (1.08-2.78)  .02

Demographics
  Age <20                            (0.71-3.77)  .25
  Aged [greater than or equal to]35  (0.91-1.78)  .16
  Black or Hispanic                  (0.37-0.82)  .00
  Divorced/separated/widowed         (0.36-1.38)  .30
  Never married                      (0.46-1.08)  .11
  Years of education                 (1.05-1.23)  .00
  Total people in household          (0.82-1.04)  .18

Income
  Current income                     (0.97-1.08)  .33
  Income data missing                (0.35-1.21)  .18

Health and pregnancy history
  First pregnancy                    (0.82-1.53)  .49
  Told to expect birth problem       (1.06-2.07)  .02
  Multiple gestation                     --       --
  Two weeks late or more                 --       --
  In good, fair, or poor health      (0.87-1.64)  .27

Other
  Year 1993                          (0.57-1.10)  .17
  Tennessee                          (0.56-1.10)  .16

Percent "yes"                           39.9


            Explanatory                      Amniocentesis
             Variables                     OR           95% CI

Health Insurance
  Tenn Care                               0.69        (0.32-1.49)
  Uninsured                               0.59        (0.12-2.85)
  Private                                 0.83        (0.44.1.57)

Demographics
  Age <20                                 2.84        (0.98-8.25)
  Aged [greater than or equal to]35       2.84        (1.82-4.44)
  Black or Hispanic                       0.71        (0.40-1.26)
  Divorced/separated/widowed              0.84        (0.32-1.26)
  Never married                           1.16        (0.62-2.17)
  Years of education                      1.03        (0.93-1.15)
  Total people in household               0.86        (0.70-1.06)

Income
  Current income                          1.00        (0.93-1.07)
  Income data missing                     0.96        (0.42-2.17)

Health and pregnancy history
  First pregnancy                         0.80        (0.50-1.27)
  Told to expect birth problem            2.80        (1.86-4.23)
  Multiple gestation                       --             --
  Two weeks late or more                   --             --
  In good, fair, or poor health           1.41        (0.92-2.17)

Other
  Year 1993                               0.93        (0.59-1.46)
  Tennessee                               0.71        (0.45-1.11)

Percent "yes"                                            13.5


            Explanatory              Amniocente
                                        sis
             Variables                P

Health Insurance
  Tenn Care                          .34
  Uninsured                          .51
  Private                            .56

Demographics
  Age <20                            .06
  Aged [greater than or equal to]35  .00
  Black or Hispanic                  .24
  Divorced/separated/widowed         .73
  Never married                      .64
  Years of education                 .53
  Total people in household          .15

Income
  Current income                     .95
  Income data missing                .91

Health and pregnancy history
  First pregnancy                    .34
  Told to expect birth problem       .00
  Multiple gestation                 --
  Two weeks late or more             --
  In good, fair, or poor health      .12

Other
  Year 1993                          .76
  Tennessee                          .13

Percent "yes"

Statistical analysis used the "proc logistic" command of SAS 6.12 with
the risk limit (/rl) option.

OR = Odds ratio, CI = confidence interval.
TABLE 4

Utilization At and After Labor/Delivery


                                                Cesarean
           Explanatory                           Section
            Variables                  OR        95% CI      P

Health insurance
 TennCare                             0.92     (0.50-1.70)  .78
 Uninsured                            1.38     (0.47-4.10)  .56
 Private                              1.27     (0.75-2.14)  .38
Demographics
 Age <20                              0.85     (0.29-2.43)  .75
 Age [greater than or equal to]35     1.54     (1.06-2.23)  .02
 Black or Hispanic                    0.85     (0.55-1.32)  .48
 Divorced/separated/widowed           1.10     (0.54-2.24)  .80
 Never married                        0.73     (0.45-1.19)  .21
 Years of education                   1.00     (0.92-1.09)  .95
 Total people in household            0.92     (0.78-1.07)  .28
Income
 Current income                       0.96     (0.90-1.02)  .16
 Income data missing                  0.78     (0.40-1.55)  .48
Health and pregnancy history
 First pregnancy                      1.06     (0.74-1.52)  .76
 Told to expect birth problem         1.61     (1.12-2.32)  .01
 Multiple gestation                   3.04     (1.30-7.07)  .01
 Two weeks late or more               1.25     (0.56-2.77)  .58
 In good, fair, or poor health        1.19     (0.84-1.69)  .34
 Low birth weight (1,500-2,500g)      1.62     (0.93-2.83)  .09
 Very low birth weight (<1,500 g)     2.42     (1.09-5.35)  .03
 Any complication after birth          -            -        -
Other
 Year 1993                            0.71     (0.49-1.02)  .06
 Tennessee                            0.78     (0.54-1.13)  .19
Percent yes                                       24.5
Percent greater than 0                              -

                                          At Labor/Delivery
                                     Regular Doctor or Affiliate
           Explanatory                     Delivered Child
            Variables                           OR

Health insurance
 TennCare                                      1.53
 Uninsured                                     0.61
 Private                                       3.56
Demographics
 Age <20                                       0.54
 Age [greater than or equal to]35              1.45
 Black or Hispanic                             0.81
 Divorced/separated/widowed                    0.92
 Never married                                 0.79
 Years of education                            1.02
 Total people in household                     0.97
Income
 Current income                                1.14
 Income data missing                           2.19
Health and pregnancy history
 First pregnancy                               0.74
 Told to expect birth problem                  0.60
 Multiple gestation                            0.72
 Two weeks late or more                        1.06
 In good, fair, or poor health                 0.96
 Low birth weight (1,500-2,500g)                 -
 Very low birth weight (<1,500 g)                -
 Any complication after birth                    -
Other
 Year 1993                                     1.23
 Tennessee                                     1.38
Percent yes
Percent greater than 0

                                      At Labor/Delivery
                                      Regular Doctor or
                                          Affiliate
           Explanatory                 Delivered Child
            Variables                95% CI      P

Health insurance
 TennCare                          (0.77-3.03)  .22
 Uninsured                         (0.21-1.78)  .37
 Private                           (1.98-6.42)  .00
Demographics
 Age <20                           (0.24-1.24)  .14
 Age [greater than or equal to]35  (0.82-2.55)  .20
 Black or Hispanic                 (0.52-1.26)  .35
 Divorced/separated/widowed        (0.40-2.12)  .85
 Never married                     (0.48-1.29)  .34
 Years of education                (0.91-1.13)  .79
 Total people in household         (0.85-1.11)  .65
Income
 Current income                    (1.02-1.70)  .02
 Income data missing               (1.01-4.76)  .05
Health and pregnancy history
 First pregnancy                   (0.49-1.11)  .14
 Told to expect birth problem      (0.38-0.93)  .02
 Multiple gestation                (0.27-1.92)  .51
 Two weeks late or more            (0.45-2.48)  .89
 In good, fair, or poor health     (0.64-1.42)  .83
 Low birth weight (1,500-2,500g)        -        -
 Very low birth weight (<1,500 g)       -        -
 Any complication after birth           -        -
Other
 Year 1993                         (0.72-2.08)  .45
 Tennessee                         (0.84-2.26)  .20
Percent yes                           83.9
Percent greater than 0                  -

                                             At Labor/Delivery
                                           Stayed 1 Day or Less
           Explanatory                          in Hospital
            Variables                        OR              95% CI

Health insurance
 TennCare                                   1.15           (0.64-2.09)
 Uninsured                                  1.16           (0.41-3.34)
 Private                                    0.95           (0.56-1.62)
Demographics
 Age <20                                    0.88           (0.36-2.13)
 Age [greater than or equal to]35           0.59           (0.39-0.88)
 Black or Hispanic                          0.73           (0.48-1.11)
 Divorced/separated/widowed                 1.32           (0.66-2.67)
 Never married                              1.21           (0.76-1.91)
 Years of education                         1.09           (0.99-1.19)
 Total people in household                  1.13           (0.98-1.29)
Income
 Current income                             1.00           (0.94-1.06)
 Income data missing                        1.46           (0.79-2.69)
Health and pregnancy history
 First pregnancy                            0.61           (0.43-0.88)
 Told to expect birth problem               0.66           (0.44-0.99)
 Multiple gestation                         0.16           (0.04-0.71)
 Two weeks late or more                     1.08           (0.49-2.42)
 In good, fair, or poor health              0.94           (0.66-1.33)
 Low birth weight (1,500-2,500g)             -                  -
 Very low birth weight (<1,500 g)            -                  -
 Any complication after birth                -                  -
Other
 Year 1993                                  0.59           (0.40-0.86)
 Tennessee                                  1.52           (1.02-2.27)
Percent yes                                                   24.3
Percent greater than 0                                          -

                                       At        After Labor/Delivery
                                   Labor/Deli
                                      very
                                    Stayed 1           Child's
                                     Day or
                                      Less
           Explanatory                 in           Hospitalization
                                    Hospital
            Variables               P                  OR

Health insurance
 TennCare                          .64                1.03
 Uninsured                         .78                0.64
 Private                           .86                0.82
Demographics
 Age <20                           .77                0.70
 Age [greater than or equal to]35  .01                0.61
 Black or Hispanic                 .14                0.45
 Divorced/separated/widowed        .43                0.66
 Never married                     .43                1.11
 Years of education                .07                1.06
 Total people in household         .09                0.92
Income
 Current income                    .99                0.90
 Income data missing               .23                1.10
Health and pregnancy history
 First pregnancy                   .01                0.78
 Told to expect birth problem      .04                1.26
 Multiple gestation                .02                1.71
 Two weeks late or more            .84                0.36
 In good, fair, or poor health     .72                1.24
 Low birth weight (1,500-2,500g)    -                 1.24
 Very low birth weight (<1,500 g)   -                 2.12
 Any complication after birth       -                 3.59
Other
 Year 1993                         .01                1.05
 Tennessee                         .04                1.13
Percent yes
Percent greater than 0

                                    After Labor/Delivery
                                           Child's
           Explanatory                 Hospitalization
            Variables                95% CI      P

Health insurance
 TennCare                          (0.52-2.03)  .94
 Uninsured                         (0.13-3.19)  .59
 Private                           (0.46-1.46)  .49
Demographics
 Age <20                           (0.21-2.41)  .57
 Age [greater than or equal to]35  (0.37-1.00)  .05
 Black or Hispanic                 (0.27-0.77)  .00
 Divorced/separated/widowed        (0.28-1.56)  .35
 Never married                     (0.66-1.86)  .71
 Years of education                (0.96-1.18)  .28
 Total people in household         (0.76-1.11)  .38
Income
 Current income                    (0.83-1.94)  .03
 Income data missing               (0.54-2.27)  .80
Health and pregnancy history
 First pregnancy                   (0.51-1.19)  .24
 Told to expect birth problem      (0.82-1.94)  .29
 Multiple gestation                (0.65-4.48)  .28
 Two weeks late or more            (0.10-1.27)  .11
 In good, fair, or poor health     (0.84-1.83)  .28
 Low birth weight (1,500-2,500g)   (0.65-2.36)  .52
 Very low birth weight (<1,500 g)  (0.90-4.96)  .08
 Any complication after birth      (2.44-5.30)  .00
Other
 Year 1993                         (0.66-1.66)  .85
 Tennessee                         (0.71-1.80)  .59
Percent yes                             -
Percent greater than 0                16.8

Statistical analysis used the "proc logistic" command of SAS 6.12 with
the risk limit (/rl) option.

OR = Odds ratio, CI = confidence interval.

References

(1.) Bonnyman G Jr: Stealth reform: market-based Medicaid in Tennessee. Health Aff 1996; 15:306-314

(2.) Bodenheimer T: The Oregon health plan: lessons for the nation. N Engl J Med 1997;337:720-723

(3.) Zuckerman S, Evans A, Holahan J: Questions for states as they turn to Medicaid managed care. Available at http://newfederalism.urban.org/html/anf_all.htm. Accessed Nov 4, 1997

(4.) Lewin VHI: The States and Private Sector: Leading Health Care Reform. Washington, DC, National Institute for Health Care Management, 1995

(5.) Gold M, Sparer M, Chu K: Medicaid managed care: lessons from five states. Health Aff 1996; 15:153-166

(6.) Schoen C, Lyons B, Rowland D, et al: Insurance matters for low-income adults: results from a five-state survey. Health Aff 1997; 16:163-171

(7.) Reforming the Health Care Payment System: State Profiles 1997. Washington, DC, American Association of Retired Persons, 1997

(8.) US Department of Commerce: Statistical Abstract of the United States, 1997. Washington, DC, Government Printing Office, 1997

(9.) Long P, Liska D: State Facts: Health Needs and Medicaid Financing. Menlo Park, Calif, The Henry J. Kaiser Family Foundation, 1998

(10.) Gruber J: The incidence of mandated maternity benefits. Am Econ Rev 1994; 84:622-641

(11.) Conover GJ, Rankin PJ, Sloan FA: Effects of Tennessee Medicaid managed care on obstetrical care and birth outcomes. J Health Polit Policy Law (In press)

(12.) Klinkman MS, Gorenflow DW, Ritsema TS: Effects of insurance coverage on quality of prenatal care. Arch Fam Med 1997; 6:557-566

(13.) Ray W, Gigante J, Mitchel E, et al: Perinatal outcomes following implementation of TennCare. JAMA 1998; 279:314-316

(14.) Miller R, Luft H: Managed care plan performance since 1980. JAMA 1994; 271:62-69

(15.) Weinstein RB, Trussell J: Declining cesarean delivery rates in California: an effect of managed care? Am J Obstet Gynecol 1998; 179:657-664

(16.) Aitken ME, Warden CR, Critchlow CW: The effect of health maintenance organization vs commercial insurance status on obstetrical management and outcome. Arch Pediatr Adolesc Med 1997; 151:1104.1108

(17.) McDuffie R Jr, Beck A, Bischoff)

Bischoff is a surname, and may refer to:

A surname

• Elmer Bischoff, visual artist
• Eric Bischoff, U.S.

 K, et al: Effect of frequency of prenatal care visits on perinatal outcome among low-risk women. JAMA 1996; 275:847-851

(18.) Levinson A, Ullman F: Medicaid managed care and infant health. J Health Econ 1998; 17:351-368

(19.) Oleske DM, Branca ML, Schmidt JB, et al: Comparison of capitated and fee-for-service Medicaid reimbursement methods on pregnancy outcomes. Health Serv Res 1998; 33:55-73

(20.) Githens PB, Glass CA, Sloan FA, et al: Maternal recall and medical records: an examination of events during pregnancy, childbirth, and early infancy. Birth 1993; 20:136-141

RECATED ARTICLE: KEY POINTS

* Initiation of care in the first trimester was particularly problematic for TennCare enrollees.

* TennCare recipients were more likely than those covered by traditional Medicaid to receive several types of tests.

* TennCare recipients were far less likely than privately insured women to have an office-based provider.

* The increased frequency of 1-day maternity stays among managed-care patients has attracted considerable attention in the media and among policy makers. In this regard, our finding of no statistically significant difference between TennCare and traditional Medicaid is interesting.

* TennCare provided some gains in terms of having access to a regular prenatal care provider. In some cases, the improved access was to a nurse rather than a physician.

From the Center for Health Policy, Law and Management, Department of Economics, and Fuqua School of Business, Duke University, Durham, NC. (May Ling Mah is a student at the University of North Carolina School of Medicine, Chapel Hill. Dr. Rankin is with Charles River Associates, Washington DC.)

Supported by funding from the Robert Wood Johnson Foundation, "Changes in Health Care Financing and Organization" program No. 028864.

Reprint requests to Frank A. Sloan, PhD, Duke University, Center for Health Policy, Law and Management, Box 90253, Durham, NC 27708.