Survey development and administration

Our goal was to survey all nonmilitary California childbirth hospitals with respect to the domains of this framework. We did not survey military hospitals or freestanding birth centers because these facilities do not routinely report administrative discharge data to the Office of Statewide Health Planning and Development, which in future work will be linked to the survey results.

We generated survey items within each domain, basing the items in part on those developed for a previous instrument that we administered to labor and delivery managers in California childbirth hospitals to document the range of policies and practices. Both interviewers (D.L.B. and D.S.F.) had considerable experience with the operations of obstetrical services, one having worked as a neonatal intensive care unit (NICU) registered nurse and the other as a clinical fellow in MFM. They were trained to assure consistent administration of the survey, which they piloted among labor and delivery managers at 5 hospitals to assure face and construct validity of the domains and individual items and to assure that nurse managers would be an adequate and reliable source of information. Items were edited in response to the pilot testing.

Upon finalization of all items, interrater reliability was assessed by having each interviewer conduct 5 interviews in the presence of the other (a total of 10 repeat interviews), with each interviewer recording results and agreement determined between the interviewers for each individual item, using Cohen’s kappa for categorical responses and the Shrout-Fleiss intraclass correlation (ICC) for continuous or mixed responses. In addition, 10 participants were retested at 3 months to assure concordance with their previous responses using the Shrout-Fleiss ICC.

Survey items that did not have good interrater or test-retest reliability (kappa < 0.8 or < 80% agreement) were eliminated. The survey contained 185 questions that resulted in 293 individual items, and it took approximately 1 hour to complete. A copy of the survey is available online ( http://amf-consulting.com/mqi/loc/ ).

Contact information for labor and delivery managers at all California childbirth hospitals was obtained through the Regional Perinatal Programs of California, which is a statewide cooperative network of public and private health care providers to promote access to high-quality health care for pregnant women and their babies. The managers were reached by phone and an appointment made for the interview. Managers were offered a $50 gift card as an incentive. A hard copy of the survey was mailed to the managers in advance to assure familiarity with the questions. If needed, multiple repeated attempts were made to contact nurse managers for participation. If no response was achieved after more than approximately 10 attempts, hospital administrators were contacted to identify and contact the most appropriate person to complete the survey. All surveys were completed between November 2012 and January 2014.

Data management and analysis

The interviewer entered information into SurveyMonkey (Palo Alto, CA), and data were exported into SAS (SAS version 9.3; SAS Institute, Cary, NC). Because of inherent differences in care provision, financing, and mission, hospitals were categorized by type (ie, integrated delivery system [IDS] hospitals [ie, hospitals that are part of a network of health care organizations that provide a continuum of health care services, with or without an obstetrical teaching service], obstetrical teaching hospitals [non-IDS], and community hospitals [all others]). Hospitals were classified as having a teaching program in obstetrics (either as a primary site or a site to which residents rotated) as reported by survey respondents, and these responses were confirmed using a national web site listing of residency programs. IDS hospitals were evaluated independently and not under their network affiliation.

Further information was added to the survey data set. The American Hospital Association (AHA) Annual Survey (2012) provided hospital ownership and rural status classification. Other information added was as follows: (1) classification as a Critical Access Hospital in California and (2) NICU level. These additional data are described in the following text.

Rural hospital status was defined by either AHA data (as having microstatus or rural status) or status as a critical access hospital. Hospitals were classified regarding whether they did or did not have an NICU. Basic/primary care hospitals (no NICU) were identified in the annual Utilization Report of Hospitals by the California Office of Statewide Health Planning and Development as having obstetrics beds but no NICU beds.

Hospitals with NICUs were classified as having an intermediate NICU, community NICU, or regional NICU as designated by California Children’s Services (CCS), which is a state program for children with special health care needs, with intermediate NICUs providing less advanced care than community or regional NICUs and regional NICUs providing the most advanced care.

Facilities with licensed NICU beds but no designation by CCS were also identified by using the annual Utilization Report of Hospitals ; such hospitals make up a fourth NICU category, NICU, no CCS. For analytical purposes, those NICUs that did not have a CCS designation were assigned a comparable designation as an intermediate NICU if they had fewer than 15 beds and a community NICU if they had 15 or more beds.

For data analysis, each survey item was described by hospital type, and differences among hospital types were tested using χ ² tests for categorical variables and nonparametric Kruskal-Wallis tests for continuous variables. Means and SDs are expressed as mean ± SD. If the response to a survey item was not obtained from a hospital, this response was classified as unknown and the denominator reduced accordingly.