Methods | Cross-sectional | Longitudinal^{*} | Overall |

Descriptive statistics† | |||

Proportion for each stage of the HIV care cascade | 276/279 (98.9%) | 20/21 (95.2%) | 296/300 (98.7%) |

Simple test statistics‡ | 78/279 (28.0%) | 6/21 (28.6%) | 84/300 (28.0%) |

Time spent in each stage of the HIV care cascade | 7/279 (2.5%) | 5/21 (23.8%) | 12/300 (4.0%) |

Transition probabilities§ | 1/279 (0.4%) | 5/21 (23.8%) | 6/300 (2.0%) |

Kaplan-Meier or cumulative incidence function | 0/279 (0%) | 5/21 (23.8%) | 5/300 (1.7%) |

Regression models† | |||

Logistic model | 101/279 (36.2%) | 7/21 (33.3%) | 108/300 (36.0%) |

Cox Proportional Hazard model | 15/279 (5.4%) | 4/21 (19.1%) | 19/300 (6.3%) |

Poisson model | 32/279 (11.5%) | 0/21 (0%) | 32/300 (10.7%) |

Competing risk model | 3/279 (1.1%) | 0/21 (0%) | 3/300 (1.0%) |

Generalised linear model | 10/279 (3.5%) | 0/21 (0%) | 10/300 (3.3%) |

Weighted and unweighted linear model | 4/279 (1.4%) | 0/21 (0%) | 4/300 (1.3%) |

Log-binomial regression model | 5/279 (1.8%) | 0/21 (0%) | 5/300 (1.7%) |

Multinomial regression model | 0/279 (0%) | 3/21 (14.3%) | 3/300 (1.0%) |

Generalised estimating equation model | 1/279 (0.4%) | 0/21 (0%) | 1/300 (0.3%) |

Other models¶ | 19/279 (6.8%) | 2/21 (9.5%) | 21/300 (7.0%) |

*Longitudinal design articles included articles that conducted both longitudinal and cross-sectional analyses.

†Percentage was estimated for each type of response and the report in this table is for yes for all binary response variables processed from multiple responses variables

‡Simple test statistics used included the χ

^{2}test, Fisher’s exact test, t-test, Wilcoxon rank -sum test and Kruskal-Wallis test.§Transition probability is a predicted probability of moving between two consecutive states in a given time.

¶Other models applied were ARIMA or SARIMA, Bayesian hierarchical models, bivariable and multivariable regression, generalised additive model, negative binomial model, generalised linear mixed model and spectrum model.

ARIMA, Autoregressive Intergrated Moving Average model; PH, Proportional Hazard; SARIMA, Seasonal Autoregressive Intergrated Moving Average model.