Reducing severe maternal morbidity for birthing persons with severe hypertension through a statewide quality improvement initiative.

BACKGROUND: Hypertensive disorders of pregnancy are a leading preventable cause of severe maternal morbidity and maternal mortality worldwide. OBJECTIVE: To assess the improvement in hospital care processes and patient outcomes associated with hypertensive disorders of pregnancy after introduction of a statewide Severe Maternal Hypertension Quality Improvement Initiative. STUDY DESIGN: A prospective cohort design comparing outcomes before and after introduction of the Illinois Perinatal Quality Collaborative statewide hypertension quality improvement initiative among 108 hospitals across Illinois. Participating hospitals recorded data for all cases of new-onset severe hypertension (>160 mm Hg systolic or >110 mm Hg diastolic) during pregnancy through 6 weeks postpartum from May 2016 to December 2017. Introduction of the statewide quality improvement initiative included implementation of severe maternal hypertension protocols, standardized patient education and discharge planning, rapid access to medications and standardized treatment order sets, and provider and nurse education. The main outcome measure was the reduction of severe maternal morbidity for pregnant/postpartum patients with severe hypertension. Key process measures include time to treatment of severe hypertension, frequency of provider/nurse debriefs, appropriate patient education, and early postpartum follow-up. RESULTS: Data were reported for 8073 cases of severe maternal hypertension. The frequency of patients with new-onset severe hypertension treated within 60 minutes increased from 41% baseline to 87% (P<.001) at the end of the initiative. The initiative was associated with increased proportion of patients receiving preeclampsia education at discharge (41% to 89%; P<.001), scheduling follow-up appointments within 10 days of discharge (68% to 83%; P<.001), and having a care team debrief after severe hypertension was diagnosed (17% to 59%; P<.001). Conversely, severe maternal morbidity was reduced from 11.5% baseline to 8.4% (P<.002) at the end of the study period. Illinois hospitals have achieved time to treatment goal regardless of hospital characteristics including geography, birth volume, and patient mix. CONCLUSION: Introduction of a statewide quality improvement effort was associated with improved time to treatment of severe hypertension and increased frequency of provider/nurse debriefs, appropriate patient education, and early postpartum follow-up scheduled at discharge, and reduced severe maternal morbidity.

Automating Ischemic Stroke Subtype Classification Using Machine Learning and Natural Language Processing.

OBJECTIVE: The manual adjudication of disease classification is time-consuming, error-prone, and limits scaling to large datasets. In ischemic stroke (IS), subtype classification is critical for management and outcome prediction. This study sought to use natural language processing of electronic health records (EHR) combined with machine learning methods to automate IS subtyping. METHODS: Among IS patients from an observational registry with TOAST subtyping adjudicated by board-certified vascular neurologists, we analyzed unstructured text-based EHR data including neurology progress notes and neuroradiology reports using natural language processing. We performed several feature selection methods to reduce the high dimensionality of the features and 5-fold cross validation to test generalizability of our methods and minimize overfitting. We used several machine learning methods and calculated the kappa values for agreement between each machine learning approach to manual adjudication. We then performed a blinded testing of the best algorithm against a held-out subset of 50 cases. RESULTS: Compared to manual classification, the best machine-based classification achieved a kappa of .25 using radiology reports alone, .57 using progress notes alone, and .57 using combined data. Kappa values varied by subtype being highest for cardioembolic (.64) and lowest for cryptogenic cases (.47). In the held-out test subset, machine-based classification agreed with rater classification in 40 of 50 cases (kappa .72). CONCLUSIONS: Automated machine learning approaches using textual data from the EHR shows agreement with manual TOAST classification. The automated pipeline, if externally validated, could enable large-scale stroke epidemiology research.

Association Between State Medical Malpractice Environment and Surgical Quality and Cost in the United States.

CONTEXT: The US medical malpractice system is designed to deter negligence and encourage quality of care through threat of liability. OBJECTIVE: To examine whether state-level malpractice environment is associated with outcomes and costs of colorectal surgery. DESIGN, SETTING, AND PATIENTS: Observational study of 116,977 Medicare fee-for-service beneficiaries who underwent colorectal surgery using administrative claims data. State-level malpractice risk was measured using mean general surgery malpractice insurance premiums; paid claims per surgeon; state tort reforms; and a composite measure. Associations between malpractice environment and postoperative outcomes and price-standardized Medicare payments were estimated using hierarchical logistic regression and generalized linear models. MAIN OUTCOME MEASURES: thirty-day postoperative mortality; complications (pneumonia, myocardial infarction, venous thromboembolism, acute renal failure, surgical site infection, postoperative sepsis, any complication); readmission; total price-standardized Medicare payments for index hospitalization and 30-day postdischarge episode-of-care. RESULTS: Few associations between measures of state malpractice risk environment and outcomes were identified. However, analyses using the composite measure showed that patients treated in states with greatest malpractice risk were more likely than those in lowest risk states to experience any complication (OR: 1.31; 95% CI: 1.22-1.41), pneumonia (OR: 1.36; 95%: CI, 1.16-1.60), myocardial infarction (OR: 1.44; 95% CI: 1.22-1.70), venous thromboembolism (OR:2.11; 95% CI: 1.70-2.61), acute renal failure (OR: 1.34; 95% CI; 1.22-1.47), and sepsis (OR: 1.38; 95% CI: 1.24-1.53; all P < 0.001). There were no consistent associations between malpractice environment and Medicare payments. CONCLUSIONS: There were no consistent associations between state-level malpractice risk and higher quality of care or Medicare payments for colorectal surgery.

Differences in empowerment and quality of life among parents of children with food allergy.

BACKGROUND: Living with food allergy has been found to adversely affect quality of life. Previous studies of the psychosocial impact of food allergy on caregivers have focused on mothers. OBJECTIVE: To describe differences in food allergy-related quality of life (FAQOL) and empowerment of mothers and fathers of a large cohort of children with food allergy. METHODS: Eight hundred seventy-six families of children with food allergy were studied. Food allergy was defined by stringent criteria, including reaction history, skin prick testing, and specific IgE. Parental empowerment and FAQOL were assessed by the adapted Family Empowerment and FAQOL-Parental Burden scales. Parental scores were compared by Wilcoxon signed rank test. Multiple regression models examined the association of parental empowerment with FAQOL. RESULTS: Mothers reported greater empowerment (P < .001) and lower FAQOL (P < .001) compared with fathers, regardless of allergen severity, type, or comorbidities. However, parental empowerment was not significantly associated with FAQOL for mothers or fathers. Although parents of children with peanut, cow milk, egg, and tree nut allergies were similarly empowered, milk and egg allergies were associated with lower FAQOL (P < .01). Parental concern in the QOL assessment was greatest for items involving fear of allergen exposure outside the home. CONCLUSION: Parental empowerment and FAQOL vary significantly among mothers and fathers of children with food allergy. Greater effects on FAQOL were seen for milk and egg compared with other food allergies. Although parents of children with food allergy might be empowered to care for their child, they continue to experience impaired FAQOL owing to fears of allergen exposure beyond their control.

Adherence with postdischarge venous thromboembolism chemoprophylaxis recommendations after colorectal cancer surgery among elderly Medicare beneficiaries.

OBJECTIVES: To assess national adherence with extended venous thromboembolism (VTE) chemoprophylaxis guideline recommendations after colorectal cancer surgery. BACKGROUND: Postoperative VTE remains a major cause of morbidity and mortality after abdominal cancer surgery. On the basis of the results from randomized controlled trials, since 2007, national guidelines have suggested that these patients be discharged on VTE chemoprophylaxis. METHODS: Medicare beneficiaries undergoing open colorectal cancer resections in 2008-2009 were identified using the Medicare Provider Analysis and Review data and limited to those who were enrolled and used Part D for their postoperative prescriptions. Postdischarge use of low-molecular-weight-heparin and other anticoagulants was assessed. RESULTS: A total of 5078 patients underwent open colorectal cancer surgery and met the inclusion criteria. Of these, 77% underwent colectomy and 23% underwent proctectomy. A prescription for an anticoagulant was filled immediately after discharge for 77 (1.5%) patients, and a low-molecular-weight-heparin for 60 (1.2%) patients. On multivariable analysis, patients were more likely to receive postdischarge VTE chemoprophylaxis if undergoing rectal cancer surgery [incidence rate ratio (IRR), 1.83; 95% confidence interval, 1.07-3.12; vs colon], if higher educational status (IRR, 2.20; 95% confidence interval, 1.23-3.95; vs low education), or if they had a higher Elixhauser comorbidity index (IRR, 1.13; 95% confidence interval, 1.01-1.25; vs lower index). CONCLUSIONS: Although VTE remains a major issue after abdominal cancer surgery, only 1.5% of Medicare beneficiaries undergoing colorectal cancer surgery received care consistent with established guidelines for postdischarge VTE chemoprophylaxis. Barriers to adherence must be elucidated to improve the quality of care for abdominal and pelvic cancer surgery patients.

Statin use and lower extremity amputation risk in nonelderly diabetic patients.

OBJECTIVE: To examine the association between use of statin and nonstatin cholesterol-lowering medications and risk of nontraumatic major lower extremity amputations (LEAs) and treatment failure (LEA or death). METHODS: A retrospective cohort of patients with Type I and Type 2 diabetes mellitus (diabetes) was followed for 5 years between 2004 and 2008. The follow-up exposure duration was divided into 90-day periods. Use of cholesterol-lowering agents, diabetic medications, hemoglobin A1c, body mass index, and systolic and diastolic blood pressures were observed in each period. Demographic factors were observed at baseline. Major risk factors of LEA including peripheral neuropathy, peripheral artery disease, and foot ulcers were observed at baseline and were updated for each period. LEA and deaths were assessed in each period and their hazard ratios (HRs) were estimated. The study took place in the U.S. Department of Veterans Affairs Healthcare system, and the subjects consisted of cholesterol drug-naïve patients with Type I or II diabetes who were treated in the U.S. Department of Veterans Affairs Healthcare system in 2003 and were <65 years old at the end of follow-up. RESULTS: Of 83,953 patients in the study cohort, 217 (0.3%) patients experienced a major LEA and 11,716 (14.0%) patients experienced an LEA or death (treatment failure) after a mean follow-up of 4.6 years. Compared with patients who did not use cholesterol-lowering agents, statin users were 35% to 43% less likely to experience an LEA (HR, 0.65; 95% confidence interval [CI], 0.42-0.99) and a treatment failure (HR, 0.57; 95% CI, 0.54-0.60). Users of other cholesterol-lowering medications were not significantly different in LEA risk (HR, 0.95; 95% CI, 0.35-2.60) but had a 41% lower risk of treatment failure (HR, 0.59; 95% CI, 0.51-0.68). CONCLUSIONS: This is the first study to report a significant association between statin use and diminished amputation risk among patients with diabetes. In this nonrandomized cohort, beneficial effects of statin therapy were similar to that seen in large-scale clinical trial experience. For LEA risk, those given nonstatins did not have a statistically significant benefit and its effect on LEA risk was much smaller compared with statins. Unanswered questions to be explored in future studies include a comparison of statins of moderate vs high potency in those with high risk of coronary heart disease and an exploration of whether the effects seen in this study are simply effects of cholesterol-lowering or possibly pleiotropic effects.

Optimizing accuracy of birth certificate data through a statewide quality improvement initiative in Illinois.

OBJECTIVE: This study examines improvement in birth certificate accuracy during a statewide quality improvement initiative. STUDY DESIGN: Participating hospitals systematically sampled 10 delivery medical records per month and compared them to corresponding birth certificates for accuracy. Accuracy was computed before implementing the initiative (Aug-Oct 2014), end of phase 1 (July 2015) and end of phase 2 (Nov-Dec 2015). Accuracy data was aggregated and compared across time points using a linear mixed model and by hospital characteristics. RESULTS: 105 hospitals participated. Birth certificate accuracy increased between baseline (89.59%) and end of phase 2 (97.00%, p < 0.001). Percent accuracy at baseline was lowest in hospitals serving at-risk populations (p < 0.01). These hospitals showed relatively greater increases in overall accuracy with no difference in accuracy by the end of the initiative. CONCLUSIONS: A statewide QI effort contributed to improvements in birth certificate accuracy. Hospitals serving at-risk populations exhibited the greatest benefit and improvement.

Effects of welfare and maternal work on recommended preventive care utilization among low-income children.

OBJECTIVES: We examined how maternal work and welfare receipt are associated with children receiving recommended pediatric preventive care services. METHODS: We identified American Academy of Pediatrics-recommended preventive care visits from medical records of children in the 1999-2004 Illinois Families Study: Child Well-Being. We used Illinois administrative data to identify whether mothers received welfare or worked during the period the visit was recommended, and we analyzed the child visit data using random-intercept logistic regressions that adjusted for child, maternal, and visit-specific characteristics. RESULTS: The 485 children (95%) meeting inclusion criteria made 41% of their recommended visits. Children were 60% more likely (adjusted odds ratios [AOR` = 1.60; 95% confidence interval [CI] = 1.27, 2.01) to make recommended visits when mothers received welfare but did not work compared with when mothers did not receive welfare and did not work. Children were 25% less likely (AOR = 0.75; 95% CI = 0.60, 0.94) to make preventive care visits during periods when mothers received welfare and worked compared with welfare only periods. CONCLUSION: The Temporary Assistance for Needy Families maternal work requirement may be a barrier to receiving recommended preventive pediatric health care.

Significant J-shaped association between body mass index (BMI) and diabetic foot ulcers.

BACKGROUND: Disagreement exists regarding the relationship between body weight and foot ulceration risk among diabetic persons. METHODS: We used a nested case-control design to estimate the association between body mass index (BMI) and 1-year and 5-year foot ulceration risk. We obtained data on all diabetic patients < 60 years of age who were treated in the US Department of Veterans Affairs healthcare system in 2003. Patient characteristics and co-morbidities were obtained at baseline. For each individual with an incident foot ulcer (case), up to four individuals were randomly selected who matched the case on age, sex, race, marital status, and calendar time. RESULTS: Crude 1-year and 5-year incidence rates were 1.35 and 6.22% after a mean follow-up of 11.8 ± 1.2 months and 55.5 ± 12.8 months, respectively. Compared with individuals with BMI 25-29.9 kg/m(2) , those with BMI 40-44.9 kg/m(2) and those with BMI ≥ 45 kg/m(2) had 25% [adjusted odds ratio (AOR) = 1.25; 95% confidence interval (CI), 1-1.56] and 83% (AOR = 1.83; 95% CI, 1.44-2.32) higher 1-year risk and 1.4 (AOR = 1.39; 95% CI, 1.26-1.54) and 2.1 (AOR = 2.08; 95% CI, 1.86-2.32) times higher 5-year risk. BMI < 25 kg/m(2) was associated with 30% higher risk at both 1 year (AOR = 1.28; 95% CI, 1.04-1.58) and 5 years (AOR = 1.27; 95% CI, 1.15-1.40). CONCLUSIONS: Our data suggest a significant J-shaped association between BMI and diabetic foot ulcers.

Prediction of 30-Day Readmission After Stroke Using Machine Learning and Natural Language Processing.

Background and Purpose: This study aims to determine whether machine learning (ML) and natural language processing (NLP) from electronic health records (EHR) improve the prediction of 30-day readmission after stroke. Methods: Among index stroke admissions between 2011 and 2016 at an academic medical center, we abstracted discrete data from the EHR on demographics, risk factors, medications, hospital complications, and discharge destination and unstructured textual data from clinician notes. Readmission was defined as any unplanned hospital admission within 30 days of discharge. We developed models to predict two separate outcomes, as follows: (1) 30-day all-cause readmission and (2) 30-day stroke readmission. We compared the performance of logistic regression with advanced ML algorithms. We used several NLP methods to generate additional features from unstructured textual reports. We evaluated the performance of prediction models using a five-fold validation and tested the best model in a held-out test dataset. Areas under the curve (AUCs) were used to compare discrimination of each model. Results: In a held-out test dataset, advanced ML methods along with NLP features out performed logistic regression for all-cause readmission (AUC, 0.64 vs. 0.58; p < 0.001) and stroke readmission prediction (AUC, 0.62 vs. 0.52; p < 0.001). Conclusion: NLP-enhanced machine learning models potentially advance our ability to predict readmission after stroke. However, further improvement is necessary before being implemented in clinical practice given the weak discrimination.

Risk of stroke after emergency department visits for neurologic complaints.

OBJECTIVE: To assess the risk of subsequent stroke among older patients discharged from an emergency department (ED) without a diagnosis of TIA or stroke. METHODS: Using electronic health record data from a large urban, university hospital and a community-based hospital, we analyzed patients aged 60-89 years discharged to home from the ED without an International Statistical Classification of Diseases and Related Health Problems, 9th or 10th Revision diagnosis of TIA or stroke. Based on the presence/absence of a head CT and the presence/absence of a chief complaint suggestive of TIA or stroke ("symptoms") during the index ED visit, we created 4 mutually exclusive groups (group 1, reference: head CT no, symptoms no; group 2: head CT no, symptoms yes; group 3: head CT yes, symptoms no; and group 4: head CT yes, symptoms yes). We calculated rates of stroke in the 30, 90, and 365 days after the index visit and used multivariable logistic regression to estimate odds ratios (ORs) for subsequent stroke. RESULTS: Among 35,622 patients (mean age 70 years, 59% women, and 16% African American), unadjusted rates of stroke in 365 days were as follows: group 4: 2.5%; group 3: 1.1%; group 2: 0.69%; and group 1: 0.54%. The adjusted OR for stroke was 3.30 (95% confidence interval [CI], 1.61-6.76) in group 4, 1.56 (95% CI, 1.16-2.09) in group 3, and 0.61 (95% CI, 0.22-1.67) in group 2. CONCLUSIONS: Among patients discharged from the ED without a diagnosis of TIA or stroke, the occurrence of a head CT and/or specific neurologic symptoms established a clinically meaningful risk gradient for subsequent stroke.

An evaluation of recurrent hypoglycemia across Chicago, Illinois.

AIMS: Recurrent hypoglycemia is understudied. This study evaluates recurrent hypoglycemia, fragmentation of care and mortality in a large urban center. METHODS: The Chicago HealthLNK Data Repository (CHDR), a de-identified electronic health record data set from institutions across Chicago, identified 9741 patients with diabetes (DM) who had hypoglycemia (emergency department (ED) or inpatient admission (IA)) from 2006 to 2012. Recurrence was defined as more than one hypoglycemia encounter, and fragmentation of health care was defined as an ED visit or IA for hypoglycemia at >1 site. RESULTS: 187,644 patients were identified with DM; of 9741 patients with hypoglycemia, 2857 (29.3%) had recurrence. Patients with ≥4 hypoglycemic encounters (n = 1035) represented 10.6%, but accounted for 40.3% hypoglycemic encounters. Of 2857 patients with recurrence, 304 patients (10.6%) had fragmented care. In those with high hypoglycemic encounters (≥4), 22% (N = 226) had ≥10 encounters; race and insurance status differences were associated with number of hypoglycemic encounters. Having hypoglycemia was associated with increased mortality compared to no hypoglycemia (n = 2696, 27.7% vs n = 20,188, 11.4%; p < 0.00001 by chi-square). CONCLUSION: A small subset of patients with hypoglycemia accounted for a large subset of hypoglycemia encounters. Targeted interventions in this high-risk, high mortality group are needed.

Surveying a New Multi-Institution Clinical Data Research Network.

Cultivated by the Patient-Centered Outcomes Research Network (PCORnet), thirteen regional clinical data research networks (CDRNs) are taking shape across the U.S. The PCORnet common data model was carefully planned, and the data marts assembled by the more than 80 data-contributing institutions (nodes) are undergoing, in 2016-2017, a series of data characterization cycles. PCORnet will adjudge each node's-and thereby, in a significant way, each CDRN's-readiness or unreadiness for multi-institution research. Certifying each node's quality and fidelity is of course essential. But in understanding network readiness there is an additional, vital dimension-one that has received too little attention. It is the development of knowledge about the nature of a CDRN's data, in its federated sense. With visualizations, how might one grasp the meta-data of a CDRN? We outline an approach that builds upon the HealthLNK Data Repository, a forerunner to the Chicago Area Patient-Centered Outcomes Research Network (CAPriCORN) CDRN.

Development of a composite measure of state-level malpractice environment.

OBJECTIVE: To develop a composite measure of state-level malpractice environment. DATA SOURCES: Public use data from the National Practitioner Data Bank, Medical Liability Monitor, the National Conference of State Legislatures, and the American Bar Association. STUDY DESIGN: Principal component analysis of state-level indicators (paid claims rate, malpractice premiums, lawyers per capita, average award size, and malpractice laws), with indirect validation of the composite using receiver-operating characteristic curves to determine how accurately the composite could identify states with high-tort activity and costs. PRINCIPAL FINDINGS: A single composite accounted for over 73 percent of total variance in the seven indicators and demonstrated reasonable criterion validity. CONCLUSION: An empirical composite measure of state-level malpractice risk may offer advantages over single indicators in measuring overall risk and may facilitate cross-state comparisons of malpractice environments.

Environmental tobacco smoke and asthma exacerbations and severity: the difference between measured and reported exposure.

OBJECTIVE: To assess the impact of measured versus reported environmental tobacco smoke (ETS) exposure on asthma severity and exacerbations in an urban paediatric population. DESIGN: We analysed cross-sectional data from the Chicago Initiative to Raise Asthma Health Equity study that followed a cohort of 561 children aged 8-14 with physician-diagnosed asthma between 2003 and 2005. Participant sociodemographic data and asthma symptoms were gathered by parental survey; exposures to ETS were determined by salivary cotinine levels and parent report. Multivariable negative binomial and ordered logistic regressions were used to assess associations between ETS and asthma outcomes. RESULTS: Among 466 children included in our analysis, 58% had moderate or severe persistent asthma; 32% had >2 exacerbations requiring a hospitalisation or an emergency room visit or same day care in the previous year. Half of caregivers reported that at least one household member smoked. In multivariable analyses, salivary cotinine was significantly associated with frequently reported exacerbations in the previous year (adjusted incidence rate ratio=1.39, 95% CI 1.09 to 1.79), but not significantly associated with asthma severity. Reported household smoking was not significantly associated with either asthma severity or frequency of exacerbations. CONCLUSIONS: Salivary cotinine was more predictive of asthma exacerbation frequency but caregiver- reported household smoking was not. Use of a nicotine biomarker may be important in both the clinical and research settings to accurately identify an important risk factor for asthma exacerbations.

Obesity paradox in amputation risk among nonelderly diabetic men.

The association between BMI and amputation risk is not currently well known. We used data for a cohort of diabetic patients treated in the US Department of Veterans Affairs Healthcare System in 2003. Men aged <65 years at the end of follow-up were examined for their amputation risk and amputation-free survival during the next 5 years (2004-2008). Compared to overweight individuals (BMI 25-29.9 kg/m(2)), the risks of amputation and treatment failure (amputation or death) were higher for patients with BMI <25 kg/m(2) and were lower for those with BMI ≥30 kg/m(2). Individuals with BMI ≥40 kg/m(2) were only half as likely to experience any (hazard ratios (HR) = 0.49; 95% confidence interval (CI), 0.30-0.80) and major amputations (HR = 0.53; 95% CI, 0.39-0.73) during follow-up as overweight individuals. While the amputation risk continued to decrease for higher BMI, amputation-free survival showed a slight upturn at BMI >40 kg/m(2). The association between obesity and amputation risk in our data shows a pattern consistent with "obesity paradox" observed in many health conditions. More research is needed to better understand pathophysiological mechanisms that may explain the paradoxical association between obesity and lower-extremity amputation (LEA) risk.

Welfare, maternal work, and on-time childhood vaccination rates.

OBJECTIVE: To examine effects of Temporary Assistance for Needy Families welfare cash assistance and maternal work requirements on "on-time" childhood vaccination rates. METHODS: A stratified random sample of Illinois children from low-income families affected by welfare reform was monitored from 1997 to 2004. Medical records from pediatricians' offices and Medicaid claims data were used to identify the timeliness of 18 recommended vaccinations. Random-intercept logistic models were used to estimate on-time vaccine administration as a function of welfare receipt and maternal work with adjustment for characteristics of the children and mothers and time-varying covariates pertaining to the administration window for each recommended vaccine dose. RESULTS: Of all recommended vaccinations, 55.9% were administered on time. On-time vaccination rates were higher when families were receiving welfare than not (57.4% vs 52.8%). Children in families that either were receiving welfare or had working mothers were 1.7 to 2.1 times more likely to receive vaccinations on time compared with children in families that were not receiving welfare and did not have working mothers. When vaccine doses were stratified according to welfare status, maternal work was associated with decreased on-time vaccination rates (odds ratio: 0.73 [95% confidence interval: 0.59-0.90]) when families received welfare but increased on-time vaccination rates (odds ratio: 1.68 [95% confidence interval: 1.27-2.22]) when they did not receive welfare. CONCLUSIONS: These results indicate that maternal work requirements of Temporary Assistance for Needy Families had negative effects on timely administration of childhood vaccinations, although receipt of welfare itself was associated with increased on-time rates.