131 6 (UN)WARRANTED VARIATION IN HOSPITAL PROTOCOLS FOR NEONATAL REFERRAL TO THE PEDIATRICIAN | 3 GOODARZI et al. anonymized by assigning numbers to the responding departments and stored in a secure, password-protected digital system at the department of Midwifery Science at the Vrije Universiteit Amsterdam. 2.2 | Data extraction We extracted criteria for neonatal referral to the pediatrician from the protocols. The variables and categories for study were selected by the research team based on previous studies on unwarranted variation in care.5 These included absence of, or variation in, the recommendations for care,7–9 variation between professional groups,14 and geographical variation.10,11 Data were extracted by the first author and a research assistant using EXCEL software.15 The following characteristics were extracted for each protocol:: type of hospital, region, type of department, date of publication, type of protocol, disciplinary collaboration, patient involvement in protocol development, and use of references. Type of hospital was categorized as: “secondary-level” (providing general obstetric and neonatal care) and “tertiary-level” (with a neonatal intensive care unit). We categorized the hospitals into five regions, based on a geographical division by the Ministry of the Interior and Kingdom Relations: “north”(Drenthe, Friesland, Groningen), “east” (Flevoland, Gelderland, Overijssel), “south”(Limburg, Noord-Brabant), “southwest” (Zuid-Holland, Zeeland), and “northwest” (Noord Nederland, Utrecht).16 Date of publication was categorized as “mentioned” (year) or “not mentioned”. Type of protocol was categorized as “hospital only” (protocol applied only to the hospital), “regional” (protocol applied to the maternity care collaboration), or “not mentioned”. A maternity care collaboration (MCC) is an alliance between care professionals in a hospital and the surrounding primary midwifery care practices, sometimes including other care professionals such as general practitioners and maternity care assistants.17 Collaboration was categorized as monodisciplinary (protocol developed only by care professionals of the obstetric or neonatal department), “multidisciplinary within hospital” (protocol developed together by care professionals from the obstetric and neonatal departments), “multidisciplinary within MCC” (protocol developed by care professionals from the MCC), and “not mentioned”. Patient involvement in protocol development was categorized as “mentioned” or “not mentioned”. References were categorized as “scientific literature”, “guidelines”, “scientific literature and guidelines”, “agreements between professionals”, “other”, and “not mentioned”. The following information was extracted from the protocols: recommendations for type of neonatal referral to the pediatrician, pediatrician's attendance at birth, location of admission, tests before referral, test criteria, timeframe of referral, screening/diagnostic tests after referral, tests’ cutoff values, treatment, discharge criteria, and referral after discharge. Type of neonatal referral was categorized as “pediatric consultation”, “neonatal admission”, and “not mentioned”. Pediatrician's attendance at birth and neonatal referral after discharge were categorized as “indicated”, “not indicated”, and “not mentioned”. Location of admission was categorized as “maternity ward”, “neonatal ward”, and “not mentioned”. Neonatal referral after discharge was categorized as “always indicated”, “only if indicated”, “not indicated”, and “not mentioned”. Information about tests before neonatal referral, test criteria, time frame of referral, screening/diagnostic tests after referral, tests” cutoff values, treatment, and discharge criteria were extracted descriptively or categorized as “not mentioned”. For all variables, the category “noncorrespondent” was used if the information within or between protocols did not correspond. The category “unclear” was used if categorization was not possible based on the description in the protocol. The category “referred to a different document” was used if we were not provided the protocol itself, but a document containing a reference to a different document. After the data extraction, we divided the indications for neonatal referral to the pediatrician into three categories: “low threshold”, “average threshold”, and “high threshold”. Indications that led to the most referrals were categorized as “low threshold”, and indications that led to the least referrals were categorized as “high threshold”. 2.3 | Analyses Statistical analyses were conducted using STATA software 14.1.18 We used descriptive statistics (n, %) to report the protocols' characteristics, content, differences between departments (interdepartment), among practices within departments (intradepartment), and between an obstetric department and a neonatal department located in the same hospital (intrahospital). We used descriptive statistics and logistic regression analyses to examine associations among referral threshold, department, and region. Univariable logistic regression analyses were conducted to calculate crude odds ratio (OR) and 95% confidence intervals (CI) for the associations between type of department and referral threshold, and region and referral threshold. Multivariable logistic regression analyses were conducted to determine adjusted ORs (aOR). The model for type of department was adjusted for region, and the model for region was adjusted for type of department as potential confounder. 1523536x, 0, Downloaded from by Vrije Universiteit Amsterdam, Wiley Online Library on [14/11/2022]. See the Terms and Conditions ( on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License