Abstract
Objectives. To evaluate the impact of the Southern Public Health Regions’ (Regions IV and IV) Collaborative Improvement and Innovation Network (CoIIN) to Reduce Infant Mortality, supported by the US Health Resources and Services Administration.
Methods. We examined pre–post change (2011–2014) for CoIIN strategies with available outcome data from vital records (early elective delivery, smoking) and the Pregnancy Risk Assessment Monitoring System (safe sleep) as well as preterm birth and infant mortality for Regions IV and VI relative to all other regions.
Results. For most outcomes, CoIIN improvements were greater in Regions IV and VI than in other regions. For example, early elective delivery decreased by 22% versus 14% in other regions, smoking cessation during pregnancy increased by 7% versus 2%, and back sleep position increased by 5% versus 2%. Preterm birth decreased by 4%, twice that observed in other regions, but infant mortality reductions did not differ significantly.
Conclusions. The CoIIN approach to public health improvement shows promise in accelerating progress in intermediate outcomes and preterm birth. Impact on infant mortality may require additional strategies and sustained efforts.
Infant mortality is a sentinel measure of population health that reflects the underlying well-being of mothers and families. Despite renewed decline since 2005, the US infant mortality rate continues to rank poorly among developed countries, with striking disparities by race/ethnicity, socioeconomic status, and geography.1–4 To address persistently high rates of infant mortality, state health officials in the 13 southern states of public health Regions IV and VI requested support from the US Department of Health and Human Services for a regional collaborative to leverage existing investments and enable state sharing of best and promising practices. Together with public and private partners, the Health Resources and Services Administration’s (HRSA’s) Maternal and Child Health Bureau launched the Regions IV and VI Collaborative Improvement and Innovation Network (CoIIN) to Reduce Infant Mortality in 2012.5 Integrating aspects of the Breakthrough Series Model of Improvement6 and the Collaborative Innovation Networks,7 CoIIN was designed to provide a platform for states to engage in collaborative learning and apply quality improvement methods to test, spread, and scale evidence-based and innovative strategies that could accelerate progress in shared priority areas within 18 to 24 months.5 Common priority or strategy areas identified by participating states were
reducing early elective deliveries,
expanding access to interconception care through Medicaid for women with previous poor pregnancy outcomes,
promoting safe infant sleep practices,
reducing smoking in pregnancy, and
improving regionalized systems of risk-appropriate perinatal care.5
Historically, CoIIN was the first multistate public health quality improvement initiative to address infant mortality. The model differs from previous federal–state partnership efforts to improve pregnancy outcomes2 in its application of the science of data-driven quality improvement and collaborative innovation at the population level. State feedback and provisional improvement results supported the utility of the CoIIN model and led to national expansion in 2014 as well as application to new topical areas and programs, including child injury, home visiting, and maternal health.8
Public health quality improvement efforts often enhance organizational or program structures and processes but are less commonly linked to health behaviors and outcomes or evaluated with an external comparison group.9,10 This analysis provides a quantitative complement to a recent implementation evaluation5 and seeks to inform future CoIIN and related public health quality improvement initiatives by examining the health outcomes of the Regions IV and VI infant mortality CoIIN. We answer the following questions using pre–post comparisons: (1) Did strategy-specific and overarching outcomes of preterm birth and infant mortality improve in Regions IV and VI and (2) were improvements in outcomes greater in Regions IV and VI than in other regions? To answer the first, we applied a simple internal control that was tracked for quality improvement, and to answer the second, we applied an external control for evaluation and attribution to CoIIN versus broader secular trends.
METHODS
Each of the 5 shared priority areas formed a strategy team, consisting of national leaders, data experts, state materrnal and child health directors and program staff, state health officers, Medicaid directors, epidemiologists, and local community or consumer partners. Strategy teams established aims, potential strategies or drivers, and common outcome measures tracked over time on a data dashboard to drive improvement through iterative plan–do–study–act cycles6 in which strategies were executed and monitored for impact and action to revise or scale them on the basis of results. CoIIN’s structure and process is described in detail elsewhere.5 Supplemental Table A summarizes the 5 team aims, strategies and drivers, primary outcome measures, and data sources. In addition to strategy team measures, CoIIN-wide progress was tracked for preterm birth and infant mortality, both overall and by race/ethnicity to monitor disparities.
As of June 2017, final outcome data were publicly available for 3 of the 5 strategy teams for the year before CoIIN started (2011) and the year of completion and subsequent national expansion (2014). Data sources for these 3 strategy teams included the birth certificate (early elective delivery and smoking cessation) and the Pregnancy Risk Assessment and Monitoring System (PRAMS; safe sleep), a state-based survey of mothers sampled from birth certificates that is supported by the Centers for Disease Control and Prevention (CDC).11 We used the period-linked birth and infant data set, which includes all calendar-year births and infant deaths linked to birth certificate information, to determine preterm birth and infant mortality outcomes. Final birth certificate data and period-linked data with state identifiers were obtained from the CDC via a request approved by the National Association for Public Health Statistics and Information Systems. PRAMS data were obtained from the CDC for states that met the response-rate threshold in any given year (65% in 2011 and 60% in 2012–2014) and via a request to the state PRAMS coordinator for all other state and year combinations that did not meet the threshold.
Measures and Study Population
Early elective delivery.
This strategy team had 2 primary outcome measures: nonmedically indicated early elective delivery and total early-term birth. The former was defined as births with an induction or cesarean delivery at 37 or 38 completed weeks of clinically estimated gestational age among all births at 37 or more completed weeks’ gestation without medical indication for early delivery. Indications captured consistently on the birth certificate include multiple birth, chronic or pregnancy-induced hypertension, eclampsia, diabetes, breech presentation, and congenital anomalies of anencephaly or meningomyelocele–spina bifida, Down syndrome, diaphragmatic hernia, omphalocele, or gastroschisis. The total early-term birth rate was defined as all deliveries at 37 and 38 weeks’ gestation among those at risk for early delivery (≥ 37 weeks’ gestation).
Sensitivity analyses that used information on cesarean birth without a trial of labor and additional indications for early delivery available for the 36 states and the District of Columbia that had implemented the 2003 birth certificate revision as of January 1, 2011, showed similar results and are not presented. Although birth certificates do not capture all indications for early delivery as defined by the Joint Commission12 and underestimate those that are captured,13–15 this bias should be considered relatively constant over time within a given state. The early-term birth rate is impervious to potential shifts in medical indications but is less specific to state strategies addressing nonindicated deliveries.
Smoking cessation.
This strategy team had 3 primary outcome measures: smoking during any trimester of pregnancy, cessation before pregnancy, and cessation during pregnancy. Cessation before pregnancy was defined as not smoking in any trimester among those who reported smoking any amount in the 3 months before pregnancy. Cessation during pregnancy was defined as not smoking in the last 3 months of pregnancy among those who reported smoking in the first 3 months of pregnancy. Only the 36 states and the District of Columbia that had implemented the recommended 2003 birth certificate revision by January 1, 2011, that captures cigarette usage by trimester of pregnancy, could be included. Michigan did not follow the standard format, and Georgia was excluded because of high rates of missing data (7.4%–21.9%) during the study period, leaving a total of 34 states and the District of Columbia (9 in Regions IV and VI).
Safe sleep.
This strategy team had 1 outcome measure available for a majority of Regions IV and VI states during the study period: back sleep position. The measure was defined as maternal report of most often placing the infant to sleep only on his or her back as opposed to side, stomach, or any other combination of positions. PRAMS sleep position data were available for 12 Regions IV and VI states and 25 states from other regions throughout the study period. Two states were not available because of delays (Hawaii) or protocol violations (Mississippi) in the 2014 files.
CoIIN-wide measures.
Overarching measures included preterm birth (< 37 completed weeks of clinically estimated gestation) and infant mortality (death at age < 1 year), as well as mortality components: neonatal mortality (death at age < 28 days), postneonatal mortality (death from age 28 days to < 1 year), sudden unexpected infant death (deaths resulting from sudden infant death syndrome, unknown cause, and accidental suffocation and strangulation in bed), and preterm-related mortality rates.16 We used 3-year rates (2010–2012 vs 2013–2015) to assess changes in infant mortality given the variability in annual state-specific rates. Infant mortality analyses were restricted to births at 22 weeks’ or more gestation to account for geographic and temporal reporting variability at earlier gestational ages.17,18
Analytic Methods
To assess improvement in outcomes targeted by CoIIN states for our first quality improvement aim, we calculated a pre–post percent change for each strategy team and CoIIN-wide measure for available Regions IV and VI states and both regions overall, comparing 2011 estimates with 2014 estimates: 
. To compare improvement in CoIIN states relative to that in other regions for our second evaluation aim, we compared percent change estimates for each Regions IV and VI state and both regions overall with percent change estimates from all other regions combined. We used relative percent change, rather than absolute differences, to control for different baseline levels across states and regions and to enable standardized comparisons across outcomes.19
We weighted all infant mortality estimates to account for the share of infant deaths by state and year that could not be linked to corresponding birth certificate information (< 5% throughout the study period). We calculated standard errors for birth20 and death21 statistics using recommended formulas. All PRAMS estimates were weighted to reflect the population, accounting for the probability of selection, nonresponse, and noncoverage. State-specific standard errors accounted for the complex survey design. Region-level standard errors that combined estimates across states, including those that did not meet CDC response-rate thresholds, could not be survey-adjusted and assumed simple random sampling using unweighted counts as a crude proxy. We evaluated statistical significance in percent change and differences in percent change relative to all other regions using 2-tailed tests ( Z ≥1.96).19
Excess rates of adverse birth outcomes among non-Hispanic Blacks contribute disproportionately to the overall Southern disadvantage in infant mortality.4 As a supplemental analysis, we examined changes in non-Hispanic Black–White disparity ratios for CoIIN-wide measures given that disparities were monitored during the initiative and 1 strategy team (safe sleep) adopted an explicit focus on disparity reduction.
RESULTS
From 2011 to 2014, the early elective delivery rate declined from 13.1% to 10.1% in Regions IV and VI, representing a 22% improvement. Of 13 states, 11 reduced their early elective delivery rates, with decreases ranging from –17% in Georgia to –44% in Louisiana. The overall improvement in Regions IV and VI and in the 11 CoIIN states with reduced rates was significantly greater than the 14% overall early elective delivery reduction in all other regions. There was also a larger decline in the total early-term birth rate in Regions IV and VI compared with other regions (–9% vs –3%; Table 1).
TABLE 1—
Early Elective Delivery Strategy Results in Regions IV and VI Compared With Other Regions: United States, 2011–2014
| Region | Early Elective Delivery Ratea |
Early Term Birth Rateb |
||||||||
| 2011, % | 2012, % | 2013, % | 2014, % | % Change | 2011, % | 2012, % | 2013, % | 2014, % | % Change | |
| Region IV | 12.2 | 11.3 | 10.1 | 9.7 | −20c,d | 31.8 | 30.9 | 29.3 | 28.9 | −9c,d |
| Alabama | 13.1 | 11.7 | 9.3 | 7.8 | −41c,d | 33.3 | 31.9 | 29.1 | 28.3 | −15c,d |
| Florida | 12.3 | 11.1 | 10.4 | 9.9 | −19c,d | 31.0 | 30.2 | 29.3 | 28.5 | −8c,d |
| Georgia | 11.7 | 11.0 | 9.9 | 9.7 | −17c,d | 32.0 | 31.2 | 29.2 | 29.1 | −9c,d |
| Kentucky | 14.6 | 13.4 | 11.3 | 11.0 | −25c,d | 34.6 | 34.3 | 31.7 | 31.4 | −9c,d |
| Mississippi | 18.9 | 18.2 | 15.4 | 13.7 | −27c,d | 41.5 | 41.5 | 37.8 | 34.8 | −16c,d |
| North Carolina | 8.0 | 8.3 | 8.1 | 8.5 | 6 | 27.1 | 26.8 | 26.3 | 26.6 | −2c |
| South Carolina | 12.9 | 10.7 | 8.9 | 8.7 | −32c,d | 32.1 | 30.4 | 28.6 | 28.6 | −11c,d |
| Tennessee | 13.6 | 12.5 | 10.7 | 10.6 | −22c,d | 32.5 | 31.3 | 29.4 | 29.3 | −10c,d |
| Region VI | 14.2 | 11.7 | 11.1 | 10.7 | −25c,d | 34.0 | 31.8 | 31.5 | 31.2 | −8c,d |
| Arkansas | 12.1 | 10.5 | 9.3 | 8.6 | −29c,d | 30.4 | 29.2 | 27.8 | 29.3 | −4c |
| Louisiana | 17.2 | 12.2 | 10.4 | 9.7 | −44c,d | 36.6 | 34.2 | 33.1 | 32.3 | −12c,d |
| New Mexico | 8.3 | 8.5 | 7.8 | 10.3 | 24 | 29.7 | 29.7 | 30.1 | 29.6 | −1 |
| Oklahoma | 14.2 | 12.8 | 11.1 | 10.6 | −25c,d | 33.8 | 32.7 | 31.1 | 30.8 | −9c,d |
| Texas | 14.5 | 11.8 | 11.6 | 11.1 | −23c,d | 34.3 | 31.8 | 31.7 | 31.3 | −9c,d |
| Regions IV and VI | 13.1 | 11.5 | 10.5 | 10.1 | −22c,d | 32.7 | 31.3 | 30.3 | 29.9 | −9c,d |
| All other regions | 8.9 | 8.3 | 7.8 | 7.6 | −14c | 27.1 | 26.7 | 26.1 | 26.1 | −3c |
Source. Birth certificate data from the National Vital Statistics System, including all 13 states in Regions IV and VI and all 37 states plus DC in other regions.
Defined as cesarean or induction at 37–38 wk of gestation among all births at 37+ weeks of gestation without medical indication for early delivery as available on the birth certificate (see Methods section for detail).
Defined as birth at 37–38 wk of gestation among all births at 37+ weeks of gestation.
Statistically significant improvement comparing 2011–2014, P < .05.
Significantly greater improvement compared with other regions, P < .05.
From 2011 to 2014, 9 CoIIN states with comparable data experienced an overall 11% reduction in any smoking during pregnancy (from 8.9% to 7.9%), as well as a 9% increase in quit rates before pregnancy (from 23.2% to 25.2%) and a 7% increase in quit rates during pregnancy (from 19.0% to 20.3%). A majority of CoIIN states (n = 7) had significant declines in smoking rates during pregnancy (range = −4% in Florida to −18% in Louisiana), 5 of 9 had significant increases in quit rates before pregnancy (range = 10% in Tennessee to 19% in Kentucky), and 5 had significant increases in quit rates during pregnancy (range = 6% in Tennessee to 16% in North Carolina). Relative to all other regions, the overall improvement in CoIIN states was significantly greater for quit rates before pregnancy (9% vs 0%) and during pregnancy (7% vs 2%) but not for any smoking in pregnancy (−11% vs −10%). However, improvement in any smoking in pregnancy was greater in Region VI relative to other regions (−15%) as a result of strong declines in 3 of 4 states with comparable data (Louisiana, Oklahoma, and Texas; Table 2).
TABLE 2—
Smoking Cessation Strategy Results in Regions IV and VI Compared With Other Regions: United States, 2011–2014
| Region | Any Smoking in Pregnancy Ratea |
Quit Rate Before Pregnancyb |
Quit Rate During Pregnancyc |
||||||||||||
| 2011, % | 2012, % | 2013, % | 2014, % | % Change | 2011, % | 2012, % | 2013, % | 2014, % | % Change | 2011, % | 2012, % | 2013, % | 2014, % | % Change | |
| Region IV | 11.4 | 11.2 | 10.9 | 10.5 | −8d | 20.5 | 22.1 | 22.6 | 22.4 | 10d,e | 17.7 | 18.2 | 17.9 | 19.2 | 9d,e |
| Alabama | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – |
| Florida | 6.7 | 6.6 | 6.6 | 6.4 | −4d | 17.5 | 19.4 | 18.4 | 16.5 | −5 | 15.0 | 15.8 | 14.9 | 16.2 | 8d,e |
| Georgia | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – |
| Kentucky | 22.5 | 22.3 | 21.6 | 20.7 | −8d | 12.9 | 16.1 | 16.8 | 15.4 | 19d,e | 11.3 | 11.9 | 11.6 | 11.5 | 2 |
| Mississippi | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – |
| North Carolina | 11.0 | 10.7 | 10.4 | 9.8 | −10d | 25.2 | 26.6 | 27.8 | 28.5 | 13d,e | 21.0 | 21.8 | 22.1 | 24.4 | 16d,e |
| South Carolina | 11.4 | 11.0 | 11.0 | 11.2 | −2 | 25.2 | 26.6 | 28.7 | 29.2 | 16d,e | 23.7 | 24.3 | 24.6 | 25.8 | 9d,e |
| Tennessee | 17.1 | 16.4 | 16.0 | 14.9 | −12d,e | 22.5 | 23.3 | 23.8 | 24.7 | 10d,e | 20.3 | 20.4 | 19.8 | 21.5 | 6d |
| Region VI | 6.3 | 6.0 | 5.8 | 5.4 | −15d,e | 27.7 | 28.4 | 28.7 | 30.1 | 8d,e | 21.5 | 21.7 | 21.6 | 22.4 | 4d |
| Arkansas | – | – | – | – | – | – | – | – | – | – | – | – | – | – | – |
| Louisiana | 9.1 | 8.6 | 8.0 | 7.4 | −18d,e | 18.0 | 18.3 | 18.5 | 19.1 | 6 | 14.9 | 14.0 | 14.7 | 14.8 | −1 |
| New Mexico | 6.3 | 7.0 | 7.2 | 6.8 | 8 | 39.9 | 36.8 | 38.3 | 37.5 | −6 | 31.8 | 29.3 | 31.2 | 30.6 | −4 |
| Oklahoma | 15.1 | 14.3 | 13.6 | 13.1 | −13d,e | 18.5 | 18.5 | 18.6 | 19.0 | 3 | 17.3 | 18.1 | 17.0 | 17.0 | −1 |
| Texas | 4.6 | 4.4 | 4.3 | 3.9 | −15d,e | 32.4 | 33.5 | 33.5 | 35.8 | 11d,e | 24.6 | 24.9 | 24.6 | 26.2 | 7d,e |
| Regions IV and VI | 8.9 | 8.6 | 8.4 | 7.9 | −11d | 23.2 | 24.4 | 24.9 | 25.2 | 9d,e | 19.0 | 19.4 | 19.2 | 20.3 | 7d,e |
| All other regions | 9.0 | 8.8 | 8.5 | 8.1 | −10d | 24.6 | 24.9 | 24.7 | 24.7 | 0 | 21.4 | 21.5 | 21.4 | 21.7 | 2d |
Note. Dash indicates data not available or not comparable throughout the study period.
Source. Birth certificate data from the National Vital Statistics System, including 34 states and DC with comparable collection of tobacco use in pregnancy; 9 of 13 states in Regions IV and VI, and 25 of 37 states and DC in all other regions (CA, CO, DC, DE, IA, ID, IL, IN, KS, MD, MO, MT, ND, NE, NH, NV, NY, OH, OR, PA, SD, UT, VT, WA, WI, WY).
Defined as smoking during any trimester.
Defined as not smoking in any trimester among those who reported smoking any amount in the 3 months before pregnancy.
Defined as not smoking in the last 3 months of pregnancy among those who reported smoking in the first 3 months of pregnancy.
Statistically significant improvement comparing 2011 with 2014, P < .05.
Significantly greater improvement compared with other regions, P < .05.
From 2011 to 2014, 11 CoIIN states with comparable data experienced an overall 5% increase in the use of the back-only sleep position (from 66.3% to 69.8%). Increases were statistically significant for 5 states (range = 8% in Oklahoma to 16% in Alabama). The overall improvement in back sleep position was significantly greater than that observed in other regions (5% vs 2%). The 4 states with improvements exceeding 10% also had significantly greater improvement relative to other regions (Table 3).
TABLE 3—
Safe Sleep Strategy Results in Regions IV and VI Compared With Other Regions: United States, 2011–2014
| Region | Back-Only Sleep Positiona |
||||
| 2011, % | 2012, % | 2013, % | 2014, % | % Change | |
| Region IV | 64.6b | 64.6 b | 67.4 b | 70.0 b | 8c,d |
| Alabama | 59.7 b | 66.5 b | 67.8 b | 69.5 | 16c,d |
| Florida | 67.2 b | 64.8 b | 65.4 b | 69.5 b | 3 |
| Georgia | 53.1 | 46.5 | 43.8 | 57.4 b | 8 |
| Kentucky | – | – | – | – | – |
| Mississippi | – | – | – | – | – |
| North Carolina | 69.4 b | 72.4 b | 75.5 b | 78.7 b | 13c,d |
| South Carolina | 71.2 b | 70.5 b | 71.0 b | 71.9 b | 1 |
| Tennessee | 68.6 b | 77.4 | 78.4 | 78.0 | 14c,d |
| Region VI | 68.2 b | 69.9 b | 70.6 b | 69.7 b | 2 |
| Arkansas | 64.2 | 64.2 | 67.2 | 68.7 b | 7 |
| Louisiana | 60.6 b | 57.6 b | 62.0 b | 63.4 b | 5 |
| New Mexico | 70.2 | 77.8 | 78.8 | 79.5 | 13c,d |
| Oklahoma | 69.9 | 72.6 | 67.8 | 75.4 | 8c |
| Texas | 69.5 b | 71.6 b | 72.2 b | 69.4 b | 0 |
| Regions IV and VI | 66.3 b | 67.1 b | 69.0 b | 69.8 b | 5c,d |
| All other regions | 76.8 b | 77.2 b | 77.5 b | 78.5 b | 2c |
Note. Dash indicates data not available or not comparable throughout the study period. All data are weighted to reflect population-based estimates accounting for the probability of selection, nonresponse, and noncoverage.
Source. Pregnancy Risk Assessment Monitoring System.
Defined as maternal report of most often placing infant to sleep only on his or her back; includes 35 states with available or comparable data; 11 of 13 states in Regions IV and VI, and 24 of 37 states in other regions (AK, CO, DE, FL, IL, ME, MD, MA, MI, MO, NE, NJ, NY, OH, OR, PA, RI, UT, VT, VA, WA, WV, WI, WY).
bEstimates include data from states that did not meet the Centers for Disease Control and Prevention response rate threshold of 65% in 2011 and 60% in 2012–2014;
Statistically significant improvement comparing 2011 with 2014, P < .05.
Significantly greater improvement compared with other regions, P < .05.
There was an overall 4% decrease in preterm birth in CoIIN states between 2011 and 2014 (from 10.9% to 10.5%). Significant decreases were noted for 9 of 13 states (range = −3% in Texas to −7% in Arkansas). The preterm birth decline was twice that observed in other regions (−2%). Infant mortality in Regions IV and VI declined by 2%, comparing 2010 to 2012 with 2013 to 2015 (from 5.8 to 5.6 per 1000). Significant reductions were observed for South Carolina (−13%) and Tennessee (−8%). However, improvements in infant mortality overall and in specific CoIIN states were not significantly greater than the 5% improvement in other regions (from 4.7 to 4.5 per 1000). Analyses of mortality components showed significant Regions IV and VI declines in postneonatal mortality (−5%) and preterm-related mortality (−4%), which were comparable to declines in other regions (Supplemental Table B). However, the decrease in sudden unexpected infant death in Tennessee was significantly greater than in other regions (−19% vs 1%). Because of slightly greater declines in preterm birth for non-Hispanic Whites compared with non-Hispanic Blacks in Regions IV and VI (5% vs 3%), there was a slight increase in the Black–White preterm birth rate ratio (from 1.41 to 1.44), but it was not significantly different than the change in other regions (from 1.46 to 1.47; Supplemental Table C). There were no significant changes in Black–White infant mortality disparities in Regions IV and VI or in other regions (Table 4).
TABLE 4—
Preterm Birth and Infant Mortality in Regions IV and VI Compared With Other Regions: United States, 2011–2014
| Region | Preterm Birth Ratea |
Infant Mortality Rate per 1000b |
||||||
| 2011, % | 2012, % | 2013, % | 2014, % | % Change | 2010–2012 | 2013–2015 | % Change | |
| Region IV | 10.9 | 10.9 | 10.7 | 10.5 | −4c,d | 6.0 | 5.9 | −2c |
| Alabama | 11.9 | 11.9 | 11.8 | 11.7 | −2 | 7.4 | 7.2 | −2 |
| Florida | 10.3 | 10.2 | 10.0 | 9.9 | −4c,d | 5.2 | 5.0 | −4 |
| Georgia | 11.0 | 10.9 | 10.7 | 10.8 | −2 | 5.7 | 6.1 | 8 |
| Kentucky | 11.3 | 11.0 | 11.0 | 10.7 | −5c | 6.0 | 5.9 | −1 |
| Mississippi | 13.5 | 13.8 | 13.1 | 12.9 | −5c | 8.2 | 8.1 | −2 |
| North Carolina | 10.2 | 10.1 | 9.9 | 9.7 | −5c,d | 5.8 | 5.7 | −1 |
| South Carolina | 11.5 | 11.3 | 11.1 | 10.8 | −6c,d | 6.5 | 5.6 | −13c |
| Tennessee | 11.1 | 11.2 | 11.1 | 10.8 | −3 | 6.6 | 6.1 | −8c |
| Region VI | 10.9 | 10.7 | 10.6 | 10.5 | −4c,d | 5.5 | 5.4 | −2 |
| Arkansas | 10.8 | 10.4 | 10.2 | 10.0 | −7c,d | 6.2 | 6.7 | 7 |
| Louisiana | 12.4 | 12.5 | 12.5 | 12.3 | −1 | 6.9 | 6.7 | −4 |
| New Mexico | 9.7 | 9.5 | 9.3 | 9.2 | −6c | 5.2 | 4.6 | −12 |
| Oklahoma | 10.8 | 10.9 | 10.6 | 10.3 | −5c | 6.5 | 6.3 | −3 |
| Texas | 10.7 | 10.5 | 10.4 | 10.3 | −3c,d | 5.1 | 4.9 | −2 |
| Regions IV and VI | 10.9 | 10.8 | 10.6 | 10.5 | −4c,d | 5.8 | 5.6 | −2c |
| All other regions | 9.3 | 9.2 | 9.1 | 9.1 | −2c | 4.7 | 4.5 | −5c |
Source. Period-linked birth and infant death file from the National Vital Statistics System, including all 13 states in Regions IV and VI and all 37 states plus DC in other regions.
Defined as birth < 37 completed weeks of clinically estimated gestation.
Defined as death < 1 y of age; restricted to births at ≥ 22 completed weeks of clinically estimated gestation.
Statistically significant improvement comparing 2011 with 2014, P < .05.
Significantly greater improvement compared with other regions, P < .05.
DISCUSSION
The results of this outcome evaluation revealed significant overall improvements from 2011 to 2014 for all outcomes targeted by the 3 CoIIN strategy teams with available final data as well as the overarching outcomes of preterm birth and infant mortality. Compared with other regions, the improvements in CoIIN states were significantly greater for most of the targeted outcomes with the exception of any smoking in pregnancy and infant mortality. CoIIN states’ achievement of even greater reductions than their counterparts indicates that the collaborative model and platform to leverage existing resources and initiatives, share strategies, and track real-time results with healthy competition can accelerate progress in a relatively short time frame.
Of all strategy-specific results, the 22% reduction in early elective delivery was most impressive. Alabama and Louisiana had the largest percentage of reductions in the country, both exceeding 40%. States in other regions achieved a smaller but still substantial 14% reduction during this time period, likely as a result of a variety of nationwide efforts including leadership and awareness campaigns such as the March of Dimes’ Healthy Babies are Worth the Wait initiative, the Centers for Medicaid and Medicare Services’ Strong Start initiative, and the Association of State and Territorial Health Officials’ Healthy Babies initiative; hospital quality reporting efforts including LeapFrog, the Joint Commission, Centers for Medicaid and Medicare Services’ inpatient quality reporting, and state-based perinatal quality collaboratives; and private and public insurance financing reform.22,23
In absolute numbers, the CoIIN results translate to an estimated 71 200 early elective deliveries averted in Regions IV and VI, of which 29 000 were potentially attributable to CoIIN and beyond predicted reductions from declines seen in other regions. The decrease in overall early-term birth rate may offer a more valid if less specific estimate of impact. An estimated 79 600 total early-term births were averted in Regions IV and VI from 2011 to 2014, of which 46 000 may be attributable to CoIIN. Although process measures were not uniformly tracked to identify effective strategies used to achieve this outcome, the most commonly selected strategy was working with hospital associations and health systems to institute “hard-stop” policies.24 In 2013, South Carolina also implemented payment reform for both Medicaid and the state’s largest private insurer, modeled in part on the 2011 Medicaid payment reform in Texas that halted payment for early elective deliveries.22,23 An evaluation of South Carolina’s efforts showed a 45% decline in early elective inductions with millions of dollars in Medicaid savings resulting from reduced neonatal intensive care admissions.25
From 2011 to 2014, the decline in any smoking in pregnancy in Regions IV and VI was similar to that in other regions (∼10%). This decline appears to mirror population-level declines in smoking achieved through tobacco control efforts as well as improved educational attainment.26 However, the 2 quit rate measures that more closely tracked the expected outcomes of strategies used by CoIIN states had significant improvements not observed in other regions. The overall 9% improvement in quit rates before pregnancy translates to 5600 fewer women smoking in pregnancy, all of which may be attributed to CoIIN given the absence of improvement noted in other regions. The 7% improvement in quitting during pregnancy translates to 1500 fewer women in CoIIN states who smoked in the last 3 months of pregnancy, of which 1100 may be attributed to CoIIN, given the 2% improvement observed in other regions. Kentucky, North Carolina, South Carolina, and Texas had improvements in quitting before pregnancy that exceeded 10%, and North Carolina’s 16% improvement in quitting during pregnancy was the highest in the country. Common strategies selected by states included promotion of quitlines to pregnant and postpartum women through social media campaigns and existing public health programs (e.g., Women, Infants, and Children [WIC], Home Visiting, Medicaid) and providing training and information to clinicians on evidence-based interventions and billing codes for reimbursement.
CoIIN states showed a 5% improvement in the safe sleep strategy measure of back-only sleep position. This improvement was higher than the 2% increase observed in other regions and translates to an estimated 80 600 more mothers (52 200 of which were potentially attributable to CoIIN) who most often placed their infants to sleep on their backs. Only 4 of 35 states in the country with available PRAMS data had improvements in back sleep position that exceeded 10% during the CoIIN time frame, and all were part of CoIIN (Alabama, North Carolina, Tennessee, New Mexico). Common safe sleep strategies included the development of strategic alliances and partnerships to endorse and promote the American Academy of Pediatrics’ safe sleep recommendations (e.g., WIC, hospitals, child care centers) and standardizing safe sleep education and training for providers. Other risk factors for sudden unexpected infant death, including soft bedding, bed sharing, and substance use,27 were not explicitly measured for this analysis. However, Tennessee did have success in significantly reducing sudden unexpected infant death by 19% (from 161.3 to 129.9 per 100 000). Tennessee has a particularly robust safe sleep program, including a public awareness campaign, partnerships with WIC and Home Visiting, and universal provider and patient education at all birthing hospitals.28
The greater overall decline in preterm birth in Regions IV and VI compared with other regions (4% vs 2%) may be the result of improved smoking quit rates during pregnancy and reductions in early elective delivery.29 An analysis by the Ohio Perinatal Quality Collaborative found that quality improvement efforts to reduce early elective deliveries also reduced the rate of late preterm births at 36 weeks.30 The absence of a differential infant mortality impact from CoIIN efforts may be understandable given the relatively short time period and the multifactorial nature of this outcome. Moreover, differential improvements in intermediate outcomes were relatively modest, with the exception of early elective delivery, which contributes more to infant morbidity than to mortality. Recent studies at hospital and state levels have not found corresponding changes in neonatal outcomes with reductions in early elective delivery.31–34 Similarly, the lack of disparity reduction in preterm birth and infant mortality may be understandable because only 1 strategy team adopted a clear disparity focus and aim. A recent analysis showed that state-level progress in reducing infant mortality disparities is possible over a longer time period.3 Future rapid-cycle efforts to reduce infant mortality and racial disparities may require more potent strategies with an explicit emphasis on disparities. Notably, the national expansion of the infant mortality CoIIN includes a focus on the social determinants of health and additional strategies including progesterone therapy and contraception.35
Limitations
This study has several limitations. First, the lack of consistently collected process measure data limits our ability to identify specific strategies linked to variations in improvement and represents an important lesson learned for subsequent iterations of CoIIN. Second, data were not universally available for certain outcomes because of variation in implementation of the revised birth certificate and PRAMS participation. Similarly, 2 strategy teams lacked outcome measure data and could not be evaluated. Continued investments in the availability, quality, and timeliness of data and surveillance systems are needed. In addition, differential outcome improvement in Regions IV and VI could have been driven by a continuation of trends that predated CoIIN. However, examination of available vital records data in preceding years indicated parallel trends between Regions IV and VI and other regions.
Public Health Implications
The infant mortality CoIIN—the first multistate public health quality improvement initiative to address birth outcomes—showed promising results in accelerating improvements in early elective delivery, smoking cessation, safe sleep, and preterm birth. CoIIN demonstrates that data-driven quality improvement and collaborative innovation can be applied to improve not only clinical care but population health interventions and outcomes. Impact on infant mortality and disparities may require additional strategies that address the social determinants of health, as well as longer term effort and follow-up. Further evaluations of the CoIIN model, including the national expansion and extensions to other topic areas, are necessary to build the evidence base for successful practices and model elements.
ACKNOWLEDGMENTS
We extend our deepest gratitude to Regions IV and VI state and local leaders and key partners on this initiative, including Abt Associates, Association of Maternal and Child Health Programs, Association of State and Territorial Health Officials, Centers for Disease Control and Prevention, Centers for Medicaid and Medicare Services, CityMatCH, Health Resources and Services Administration, and March of Dimes. We specifically thank the strategy team leaders, data and methods experts, and staffers from these states and organizations (presented in alphabetical order): Michelle Alletto, Sarah Ball, Wanda Barfield, Al Brann, Dabo Brantley, Stephen Cha, Lekisha Daniel-Robinson, Suzanna Dooley, Brent Ewig, Katherine Flaherty, Rebekah Gee, Carol Gilbert, David Goodman, Paul Halverson, Norm Hess, Kay Johnson, Laurin Kasehagen, Lyn Kieltyka, Deborah Klein Walker, David Lakey, Kate Menard, Erin Reiney, Cheryl Robbins, Morrisa Rice, Ellen Schliecher Pliska, Carrie Shapiro-Mendoza, Ruth Ann Shepherd, Caroline Stampfel, Cathy Taylor, Debra Wagler, Kathy Watters, Meghan Woo, and Kim Wyche Etheridge. We also thank the Vital Statistics Cooperative Program and the Pregnancy Risk Assessment and Monitoring System Working Group for coordinating the collection of data used in this evaluation.
Note. The findings and conclusions in this article are those of the authors and do not necessarily represent the official position of the Health Resources and Services Administration.
HUMAN PARTICIPANT PROTECTION
Given the use of publicly available data without individual identifiers, this analysis was not subject to institutional review board approval.
REFERENCES
- 1.Mathews TJ, Driscoll AK. Trends in Infant Mortality in the United States, 2005-2014. Atlanta, GA: National Center for Health Statistics; 2017. NCHS Data Brief No. 279. [PubMed]
- 2.Heisler EJ. The US Infant Mortality Rate: International Comparisons, Underlying Factors, and Federal Programs. Washington, DC: Congressional Research Service; 2012. [Google Scholar]
- 3.Brown Speights JS, Goldfarb SS, Wells BA, Beitsch L, Levine RS, Rust G. State-level progress in reducing the Black–White infant mortality gap, United States, 1999–2013. Am J Public Health. 2017;107(5):775–782. doi: 10.2105/AJPH.2017.303689. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hirai AH, Sappenfield WM, Kogan MD et al. Contributors to excess infant mortality in the US South. Am J Prev Med. 2014;46(3):219–227. doi: 10.1016/j.amepre.2013.12.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ghandour RM, Flaherty K, Hirai A, Lee V, Walker DK, Lu MC. Applying collaborative learning and quality improvement to public health: Lessons from the Collaborative Improvement and Innovation Network (CoIIN) to reduce infant mortality. Matern Child Health J. 2017;21(6):1318–1326. doi: 10.1007/s10995-016-2235-2. [DOI] [PubMed] [Google Scholar]
- 6.The Breakthrough Series: IHI’s Collaborative Model for Achieving Breakthrough Improvement. Boston, MA: Institute for Healthcare Improvement; 2003. IHI White Paper Series. [Google Scholar]
- 7.Gloor PA. Swarm Creativity: Competitive Advantage Through Collaborative Innovation Networks. New York, NY: Oxford University Press; 2006. [Google Scholar]
- 8.Maternal and Child Health Bureau. Collaborative Improvement and Innovation Networks. Rockville, MD: Health Resources and Services Administration. Available at: https://mchb.hrsa.gov/maternal-child-health-initiatives/collaborative-improvement-innovation-networks-coiins. Accessed September 12, 2017.
- 9.Dilley JA, Bekemeier B, Harris JR. Quality improvement interventions in public health systems: a systematic review. Am J Prev Med. 2012;42(5, suppl 1):S58–S71. doi: 10.1016/j.amepre.2012.01.022. [DOI] [PubMed] [Google Scholar]
- 10.McLees AW, Nawaz S, Thomas C, Young A. Defining and assessing quality improvement outcomes: a framework for public health. Am J Public Health. 2015;105(suppl 2):S167–S173. doi: 10.2105/AJPH.2014.302533. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Centers for Disease Control and Prevention. Pregnancy Risk Assessment Monitoring System (PRAMS). Available at: https://www.cdc.gov/prams/index.htm. Accessed January 8, 2018.
- 12.Joint Commission. Specifications for Joint Commission National Quality Measures (v2015B). Available at: https://manual.jointcommission.org/releases/TJC2015B/MIF0166.html. Accessed January 8, 2018.
- 13.Dietz P, Bombard J, Mulready-Ward C et al. Validation of selected items on the 2003 US standard certificate of live birth: New York City and Vermont. Public Health Rep. 2015;130(1):60–70. doi: 10.1177/003335491513000108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Martin JA, Wilson EC, Osterman MJ, Saadi EW, Sutton SR, Hamilton BE. Assessing the quality of medical and health data from the 2003 birth certificate revision: results from two states. Natl Vital Stat Rep. 2013;62(2):1–19. [PubMed] [Google Scholar]
- 15.Northam S, Knapp TR. The reliability and validity of birth certificates. J Obstet Gynecol Neonatal Nurs. 2006;35(1):3–12. doi: 10.1111/j.1552-6909.2006.00016.x. [DOI] [PubMed] [Google Scholar]
- 16.Callaghan WM, MacDorman MF, Rasmussen SA, Qin C, Lackritz EM. The contribution of preterm birth to infant mortality rates in the United States. Pediatrics. 2006;118(4):1566–1573. doi: 10.1542/peds.2006-0860. [DOI] [PubMed] [Google Scholar]
- 17.Ehrenthal DB, Wingate MS, Kirby RS. Variation by state in outcomes classification for deliveries less than 500 g in the United States. Matern Child Health J. 2011;15(1):42–48. doi: 10.1007/s10995-010-0566-y. [DOI] [PubMed] [Google Scholar]
- 18.Goyal NK, DeFranco E, Kamath-Rayne BD, Beck AF, Hall ES. County-level variation in infant mortality reporting at early previable gestational ages. Paediatr Perinat Epidemiol. 2017;31(5):385–391. doi: 10.1111/ppe.12376. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Keppel KG, Pearcy JN, Klein RJ. Measuring progress in Healthy People 2010. Healthy People 2010 Stat Notes. 2004; (25):1–16. [PubMed]
- 20.National Center for Health Statistics. User guide to the 2010 Natality Public Use File. Available at: https://ftp.cdc.gov/pub/Health_Statistics/NCHS/Dataset_Documentation/DVS/natality/UserGuide2010.pdf. Accessed September 12, 2017.
- 21.Matthews TJ, MacDorman MF, Thoma ME. Infant mortality statistics from the 2013 period linked birth/infant death data set. Natl Vital Stat Rep. 2015;64(9):1–30. [PubMed] [Google Scholar]
- 22.Jensen JR, White WM, Coddington CC. Maternal and neonatal complications of elective early-term deliveries. Mayo Clin Proc. 2013;88(11):1312–1317. doi: 10.1016/j.mayocp.2013.07.009. [DOI] [PubMed] [Google Scholar]
- 23.Association of State and Territorial Health Officials. Early elective delivery issue brief. Arlington, VA: Association of State and Territorial Health Officials. Available at: http://www.astho.org/Early-Elective-Delivery-Issue-Brief. Accessed August 14, 2017.
- 24.Clark SL, Frye DR, Meyers JA et al. Reduction in elective delivery at <39 weeks of gestation: comparative effectiveness of 3 approaches to change and the impact on neonatal intensive care admission and stillbirth. Am J Obstet Gynecol. 2010;203(5):449.e1–449.e6. doi: 10.1016/j.ajog.2010.05.036. [DOI] [PubMed] [Google Scholar]
- 25.Perelman N, Delbanco S, Vargas-Johnson A. Using Education, Collaboration, and Payment Reform to Reduce Early Elective Deliveries: A Case Study of South Carolina’s Birth Outcomes Initiative. Berkeley, CA: Catalyst for Payment Reform; 2013. [Google Scholar]
- 26.Jamal A, King BA, Neff LJ, Whitmill J, Babb SD, Graffunder CM. Current cigarette smoking among adults—United States, 2005–2015. MMWR Morb Mortal Wkly Rep. 2016;65(44):1205–1211. doi: 10.15585/mmwr.mm6544a2. [DOI] [PubMed] [Google Scholar]
- 27.American Academy of Pediatrics. SIDS and other sleep-related infant deaths: updated 2016 recommendations for a safe infant sleeping environment. Pediatrics. 2016;138(5) doi: 10.1542/peds.2016-2938. [DOI] [PubMed] [Google Scholar]
- 28.Association of State and Territorial Health Officials. SIDS/SUID/Safe Sleep Roundtable Report. Arlington, VA: Association of State and Territorial Health Officials. Available at: http://www.astho.org/Maternal-and-Child-Health/Safe-Sleep/ASTHO-Safe-Sleep-Roundtable-Report. Accessed September 20, 2017.
- 29.Schoen CN, Tabbah S, Iams JD, Caughey AB, Berghella V. Why the United States preterm birth rate is declining. Am J Obstet Gynecol. 2015;213(2):175–180. doi: 10.1016/j.ajog.2014.12.011. [DOI] [PubMed] [Google Scholar]
- 30.Donovan EF, Lannon C, Bailit J, Rose B, Iams JD, Byczkowski T. A statewide initiative to reduce inappropriate scheduled births at 36(0/7)-38(6/7) weeks’ gestation. Am J Obstet Gynecol. 2010;202(3):243.e1–243.e8. [DOI] [PubMed]
- 31.Howell EA, Zeitlin J, Hebert PL, Balbierz A, Egorova N. Association between hospital-level obstetric quality indicators and maternal and neonatal morbidity. JAMA. 2014;312(15):1531–1541. doi: 10.1001/jama.2014.13381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Hutcheon JA, Strumpf EC, Harper S, Giesbrecht E. Maternal and neonatal outcomes after implementation of a hospital policy to limit low-risk planned caesarean deliveries before 39 weeks of gestation: an interrupted time-series analysis. BJOG. 2015;122(9):1200–1206. doi: 10.1111/1471-0528.13396. [DOI] [PubMed] [Google Scholar]
- 33.Little SE, Robinson JN, Puopolo KM et al. The effect of obstetric practice change to reduce early term delivery on perinatal outcome. J Perinatol. 2014;34(3):176–180. doi: 10.1038/jp.2013.166. [DOI] [PubMed] [Google Scholar]
- 34.Snowden JM, Muoto I, Darney BG et al. Oregon’s hard-stop policy limiting elective early-term deliveries: association with obstetric procedure use and health outcomes. Obstet Gynecol. 2016;128(6):1389–1396. doi: 10.1097/AOG.0000000000001737. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.National Institute for Children’s Health Quality. Collaborative Improvement and Innovation Network to Reduce Infant Mortality (IM CoIIN). Boston, MA: National Institute for Children’s Health Quality. Available at: http://nichq.org/project/collaborative-improvement-and-innovation-network-reduce-infant-mortality-im-coiin. Accessed September 7, 2017.