Skip to main content

Table 4 Generalised linear model (GLM) results

From: The impact of socioeconomic factors on the healthcare costs of people living with HIV in Turkey

 (1)a  (2)
VARIABLES/LINKbPOWER (1.3)%95 Confidence Interval Incremental Effects
CD4 (200–499 cells/mm3)−14,556.84***−15,199.1−13,914.6− 898.1***
 − 327.678  −19.83
CD4 (≥ 500 cells/mm3)−16,868.45***−17,503.1− 16,233.8−1046***
 −323.80  −19.59
Wealth (≥ min wage)−21,868.29***−22,411.3−21,325.3− 1365***
 − 277.05  −17.08
Wealth (≥ 2x min wage)−20,391.69***−20,940.9−19,842.5− 1268***
 − 280.21  −17.24
Employment (employed)22,637.573***22,206.923,068.251482***
 −219.74  −14.93
Education (high school & university)911.83***512.041311.6258.24***
 −203.98  −13.04
Alcohol (yes)1439.45***996.981881.9291.65***
 − 225.75  −14.34
Smoke (yes)7150.11***6780.947519.27458.0***
 −188.35  −12.12
Psychiatric drug (yes)10,415.17***9844.4210,985.91651.3***
 −291.20  −17.87
Drug addiction (yes)8225.30***7755.718694.90518.1***
 − 239.59  −14.91
Sexual orientation (heterosexual)− 575.12***− 983.17−167.06−36.69***
 −208.20  −13.28
Age (40–54)14,638.10***14,169.6715,106.53932.0***
 − 238.10  −15
Age (≥ 55)31,475.19***30,741.6532,208.731934***
 −374.26  −22.18
Gender (female)13,398.90***12,782.6814,015.13835.1***
 − 314.41  −19.18
Marriage status (married)10,222.66***9720.72110,724.59643.6***
 − 256.09  −15.92
Child (yes)6337.71***5809.1236866.289401.4***
 − 269.69  −16.97
Time diagnosed c (≥ 10 years)−11,356.30***−11,967.4− 10,745.2− 742.19***
 −311.79  73.37
Constant46,693.04***45,804.7147,581.384225***
 − 453.24  −5.254
Observations153  153
AIC1066   
BIC160,793   
Log-likelihood−81,495   
Deviance161,472   
  1. Abbreviations: Min wage minimum wage, CD4 CD4 T-cell count, AIC Akaike information criterion, BIC Bayesian information criterion
  2. a *** p < 0.01
  3. b Parentheses show the categories of the independent variables that are included in the estimations. Reference categories are omitted
  4. c Margins command in STATA is unable to produce incremental effects for the variable Timediagnosed. We therefore calculated it by hand following Williams [22]