Table 1.
Summary characteristics of the 23 studies included in this review.
| Author | Country of study | Year of publication (data) | Study design | Type of analysis | Exclusions in study | Study population | Number of patients | Statistical methods | Study quality |
|---|---|---|---|---|---|---|---|---|---|
| Bender et al. (18) | America | 2013 (1999 and 2002) | Retrospective | Multivariable (prediction model) | (1) Died prior to NICU; (2) admission for preterminal comfort care; (3) congenital anomaly | All gestations | 908 | Log-logistic model and Weibull model | Moderate |
| Berry et al. (28) | Canada | 2008 (2002) | Retrospective | Multivariable | Admissions for convalescent or step-down care | All gestations | 625 | Logistic regression | Moderate |
| Charlton et al. (35) | America | 2019 (2014) | Retrospective | Multivariable | (1) No IVFs for at least 48 h; (2) age >7 days at NICU admission; (3) congenital heart disease with surgery at <7 days of life; (4) congenital anomaly; (5) NICU deaths; (6) severe kidney anomaly | Newborns with intravenous fluids longer than 48 h | 2,110 | Linear regression | Moderate |
| Ge et al. (1) | China | 2022 (2016–2021) | Retrospective | Multivariable | (1) congenital anomaly; (2) transfer; (3) NICU deaths; (4) discharge against medical advice | 28–31 weeks gestational age | 709 | Two-piece-wise linear regression model | High |
| Hintz et al. (9) | America | 2010 (2002–2005) | Retrospective | Multivariable (prediction model) | (1) chromosomal or congenital anomalies; (2) In hospital one year; (3) transfer; (4) missing data; (5) NICU deaths | <27 weeks gestational age | 2,254 | Linear and logistic regression models | High |
| Jetton et al. (29) | America | 2017 (2014) | Retrospective | Multivariable | (1) No IVFs for at least 48 h; (2) age >14 days at NICU admission; (3) congenital heart disease with surgery at <7 days of life; (4) lethal chromosomal anomaly; (5) mortality <48 h; (6) severe, bilateral congenital renal disease | Newborns with intravenous fluids longer than 48 h | 2,022 | Linear regression | High |
| Kheiry et al. (43) | Iran | 2020 (2016–2018) | Retrospective | Multivariable (prediction model) | Not mentioned | All gestations | 603 | Cox model and four parametric models | Moderate |
| Klinger et al. (37) | Israel | 2005 (1995–2002) | Retrospective | Multivariable | NICU deaths | <1,500 g birth weight | 9,310 | Logistic regression | High |
| Lee et al. (27) | America | 2013 (2008–2010) | Prospective | Multivariable (prediction model) | (1) NICU deaths; (2) transfer; (3) congenital anomaly; (4) surgeries | 401–1,000 g birth weight | 2,012 | Linear mixed model | High |
| Lee et al. (34) | America | 2016 (2008–2011) | Retrospective | Multivariable (prediction model) | (1) congenital anomaly; (2) readmission; (3) NICU deaths; (4) surgeries; (5) weight gain of 65 g per day | 401–1,500 g birth weight; 22–29 weeks gestational age; other larger and older infants | 23,551 | Negative binomial Generalized linear model | Moderate |
| Eken et al. (30) | Turkey | 2016 (2012–2014) | Retrospective | Multivariable | (1) NICU deaths; (2) transfer; (3) missing data | All gestations | 3,607 | Linear regression | Moderate |
| Eken et al. (36) | Turkey | 2017 (2013–2015) | Retrospective | Multivariable | (1) Missing data; (2) transfer; (3) twin pregnancies; (4) congenital anomaly | 24–34 weeks of gestation following PPROM | 331 | Linear regression | Moderate |
| Murki et al. (24) | India | 2020 (2016–2018) | Prospective | Multivariable | (1) Age >1 days at NICU admission; (2) readmission; (3) NICU deaths; (4) transfer | 25–33 weeks gestational age | 3,095 | Linear regression | Moderate |
| Patil et al. (38) | America | 2021 (1997–2018) | Retrospective | Multivariable | (1) Congenital anomaly; (2) chromosomal abnormalities/aneuploidies | 22–36 weeks gestational age | 760,037 | Logistic regression | Moderate |
| Payne et al. (39) | America | 2004 (1998–1999) | Retrospective | Multivariable | (1) NICU deaths; (2) transfer; (3) early-onset sepsis; (4) congenital anomaly; (5) missing data | <1,500 g birth Weight | 2,809 | Logistic regression | Moderate |
| Pepler et al. (42) | South Africa | 2012 (2007–2008) | Retrospective | Multivariable (prediction model) | (1) NICU deaths; (2) transfer; (3) missing data | (1) 401–1,500 g birth weight; (2) 22–29 weeks gestational age (3) Age <28 days at NICU admission | 3,703 | Log-linear models | Moderate |
| Różańska et al. (25) | Poland | 2015 (2009–2013) | Prospective | Multivariable | (1) Missing data; (2) NICU deaths | <1,500 g birth weight | 2,003 | Generalized linear model | Moderate |
| Sahiledengle et al. (26) | Ethiopia | 2020 (2018–2019) | Prospective | Multivariable | (1) Died prior to NICU; (2) congenital anomaly | All gestations | 191 | Cox regression | Moderate |
| Shah et al. (40) | Eritrea | 2012 (2006) | Retrospective | Multivariable | NICU deaths | All gestations | 953 | Linear regression | Moderate |
| Singh et al. (31) | India | 2021 (2018–2019) | Retrospective | Multivariable (prediction model) | (1) Congenital anomaly; (2) palliative care; (3) request for discharge; (4) transfer; (5) NICU deaths | All gestations | 1,047 | Log model | Moderate |
| Xie et al. (32) | China | 2022 (2012–2020) | Retrospective | Multivariable | (1) Missing data; (2) readmission; (3) hospitalization for <1 day | All gestations | 16,094 | Cox regression | Moderate |
| Ye et al. (41) | China | 2011 (2006–2009) | Retrospective | Multivariable | (1) NICU deaths; (2) transfer | 24–34 weeks of gestation following PPROM | 289 | Linear regression | Moderate |
| Zhang et al. (33) | China | 2022 (2019) | Retrospective | Multivariable | (1) Congenital anomaly; (2) missing data; (3) NICU deaths; (4) discharge against medical advice; (5) transfer to nonparticipating hospitals | <32 weeks gestational age | 6,580 | Linear regression | Moderate |
NICU, neonatal intensive care unit.