1 Background
-
Which covariates are the most determinant factors for LBW?
-
Which fitted model for the birth weight is statistically plausible?
-
Is there a significant within and between regional heterogeneity of weights of child at birth?
2 Methods
2.1 Data Source
2.2 Study Population
2.3 Variables of the Study
2.3.1 Response Variable
Variable | Presentation of variable | Factor coding |
---|---|---|
Child weight at birth | Child weight | 1 = Small birth weight (< 2500 g) |
0 = Large birth weight (≤ 2500 g) |
2.3.2 Predictor (Explanatory Variables)
Covariates | Categories |
---|---|
Sex | 0 = female, 1 = male |
Wealth status | 1 = poor, 2 = middle, 3 = rich |
Residence | 0 = rural, 1 = urban |
Age | 1 = 15–19, 2 = 20–39, 3 = 40–49 |
Terminated pregnancy | 0 = no, 1 = yes |
Antenatal visit | 1 = no visit, 2 = 1–4, 3 = \(\ge \,5\) |
Marital Status | 1 = married, 2 = widowed, 3 = divorced |
Vaccination | 0 = no, 1 = yes |
Anemia | 1 = not anemic, 2 = moderate, 3 = sever |
Mother education | 1 = no education, 2 = primary, 3 = secondary and above |
Birth order | 1 = 1–4, 2 = 5–9, \(3=\ge 10\) |
Preceding birth interval | 1 = 1–5, 2 = 6–10, \(3=\ge 11\) |
2.4 Method of Data Analysis
2.4.1 Marginal Models
2.4.2 Generalized Estimating Equations (GEE)
2.4.3 Parameter Estimation for GEE
2.4.4 Alternating Logistic Regression (ALR)
2.4.5 Parameter Estimation for ALR
2.5 Cluster Specific Models
2.5.1 Generalized Linear Mixed Model (GLMM)
2.5.2 Parameter Estimation for GLMM
2.6 Model Building
3 Results
Covariates | Child weight (%) | Total | |
---|---|---|---|
Large weight (≤ 2500 g) | Small weight (< 2500 g) | ||
Mother’s age at pregnancy | |||
15–19 | 36 (26.7) | 99 (73.3) | 135 |
20–39 | 1795 (67.5) | 866 (32.5) | 2661 |
40–49 | 271 (63.2) | 158 (36.8) | 429 |
Mother’s education | |||
No education | 1567 (62.9) | 923 (37.1) | 2490 |
Primary | 484 (72.0) | 188 (28.0) | 67 |
Secondary and above | 51 (81.0) | 12 (19.0) | 63 |
Mother’s marital status | |||
Married | 1924 (66.6) | 963 (33.4) | 2887 |
Widowed | 52 (65.0) | 28 (35.0) | 80 |
Divorced | 126 (48.8) | 132 (51.2) | 258 |
Number of ANC visits | |||
No ANC visit | 1331 (62.2) | 809 (37.8) | 2140 |
1–4 | 575 (69.6) | 251 (30.4) | 826 |
≤ 5 | 196 (75.7) | 63 (24.3) | 259 |
Ever had vaccination | |||
Yes | 1681 (68.8) | 764 (31.2) | 2445 |
No | 421 (54.0) | 359 (46.0) | 780 |
Ever had terminated pregnancy | |||
Yes | 411 (59.5) | 280 (40.5) | 691 |
No | 1691 (66.7) | 843 (33.3) | 2534 |
Maternal anemia | |||
Not anemic | 1011 (71.1) | 411 (28.9) | 1422 |
Moderate | 997 (63.7) | 567 (36.3) | 1564 |
Sever | 94 (39.3) | 145 (60.7) | 239 |
Sex of child | |||
Female | 972 (60.9) | 625 (39.2) | 1597 |
Male | 1130 (69.4) | 498 (30.6) | 1628 |
Birth order | |||
1–4 | 1071 (66.9) | 585 (33.1) | 1656 |
5–9 | 912 (65.6) | 479 (34.4) | 1391 |
\(\ge\)10 | 119 (64.7) | 59 (35.3) | 178 |
Preceding birth interval | |||
1–5 | 1835 (65.3) | 975 (34.7) | 2810 |
6–10 | 193 (64.3) | 107 (35.7) | 300 |
\(\ge\)11 | 74 (64.3) | 41 (35.7) | 115 |
Wealth Status | |||
Poor | 1127 (61.2) | 714 (38.8) | 1841 |
Middle | 349 (64.7) | 190 (35.3) | 539 |
Rich | 626 (74.1) | 219 (25.9) | 845 |
Residence | |||
Rural | 1875 (64.7) | 1025 (35.3) | 2900 |
Urban | 227 (69.8) | 98 (30.2) | 325 |
Total | 2102 (65.2) | 1123 (34.8) | 3225 |
3.1 Analysis of Marginal Models
3.1.1 Analysis of Generalized Estimating Equations (GEE)
Coeff. | Estimates | Exchangeable | Estimates | Independent | ||
---|---|---|---|---|---|---|
Model based (SE) | Empirical (SE) | Model based (SE) | Empirical (SE) | |||
\(\beta _{0}\) | 0.6468 | 0.2481 | 0.2523 | 0.7185 | 0.2300 | 0.2820 |
\(\beta _{1}\) | − 0.3449 | 0.0756 | 0.0779 | − 0.3536 | 0.0769 | 0.0801 |
\(\beta _{2}\) | 0.0249 | 0.1064 | 0.1045 | − 0.3536 | 0.1065 | 0.1114 |
\(\beta _{3}\) | − 0.3505 | 0.1037 | 0.1013 | − 0.3887 | 0.0994 | 0.1039 |
\(\beta _{4}\) | − 0.9798 | 0.2359 | 0.2532 | − 1.0449 | 0.2190 | 0.2780 |
\(\beta _{5}\) | − 0.8337 | 0.2542 | 0.2677 | − 0.8565 | 0.2395 | 0.2914 |
\(\beta _{6}\) | − 0.1361 | 0.0929 | 0.0985 | − 0.1307 | 0.0933 | 0.0996 |
\(\beta _{7}\) | − 0.2818 | 0.1635 | 0.1538 | − 0.2426 | 0.1633 | 0.1554 |
\(\beta _{8}\) | 0.3448 | 0.1480 | 0.1674 | 0.4387 | 0.1451 | 0.1908 |
\(\beta _{9}\) | 0.0198 | 0.2403 | 0.2411 | 0.0698 | 0.2423 | 0.2320 |
\(\beta _{10}\) | − 0.2584 | 0.0962 | 0.1020 | − 0.2930 | 0.0931 | 0.1088 |
\(\beta _{11}\) | 0.2262 | 0.0811 | 0.0854 | 0.2322 | 0.0814 | 0.0872 |
\(\beta _{12}\) | 0.6872 | 0.1671 | 0.1910 | 0.8098 | 0.1620 | 0.2226 |
\(\beta _{13}\) | − 0.1951 | 0.1026 | 0.1059 | − 0.1783 | 0.1015 | 0.1111 |
\(\beta _{14}\) | − 0.3240 | 0.3351 | 0.3337 | − 0.3595 | 0.3402 | 0.3461 |
Covariates | Analysis of GEE parameter estimates empirical standard error estimates | ||||||
---|---|---|---|---|---|---|---|
Level | Estimate | SE | 95% CI | Z | Pr \(>|Z|\) | ||
Intercept | 0.6468 | 0.2523 | 0.1523 | 1.1414 | 2.56 | 0.0104* | |
Sex | Male | − 0.3449 | 0.0779 | − 0.4976 | − 0.1923 | − 4.43 | < 0.0001* |
Wealth index | Middle | 0.0249 | 0.1045 | − 0.1800 | 0.2298 | 0.24 | 0.8118 |
Rich | − 0.3505 | 0.1013 | − 0.5491 | − 0.1520 | − 3.46 | 0.0005* | |
Age of mothers | 20–39 | − 0.9798 | 0.2532 | − 1.4761 | − 0.4835 | − 3.87 | 0.0001* |
40–49 | − 0.8337 | 0.2677 | − 1.3585 | − 0.3090 | − 3.11 | 0.0018* | |
Antenatal care | 1–4 | − 0.1361 | 0.0985 | − 0.3292 | 0.0571 | − 1.38 | 0.1674 |
Five and above | − 0.2818 | 0.1538 | − 0.5831 | − 0.1196 | − 3.83 | 0.0069* | |
Marital status | Divorced | 0.3448 | 0.1674 | 0.0167 | 0.6728 | 2.06 | 0.0394* |
Widowed | 0.0198 | 0.2411 | − 0.4526 | 0.4923 | 0.08 | 0.9345 | |
Vaccination | Yes | − 0.2584 | 0.1020 | − 0.4584 | − 0.0584 | − 2.53 | 0.0113* |
Anemia | Moderate | 0.2262 | 0.0854 | 0.0588 | 0.3936 | 2.65 | 0.0081* |
Sever | 0.6872 | 0.1910 | 0.3129 | 1.0614 | 3.60 | 0.0003* | |
Education level | Primary | − 0.1951 | 0.1059 | − 0.4027 | − 0.1126 | − 3.84 | 0.0056* |
Secondary and above | − 0.3240 | 0.3337 | − 0.9780 | 0.3299 | − 0.97 | 0.3314 | |
QIC = 3986.4033 | |||||||
Alpha (\(\alpha\)) = 0.0857 |
3.1.2 Analysis of Alternating Logistic Regression Model (ALR)
Covariates | Analysis of ALR parameter estimates empirical standard error estimates | ||||||
---|---|---|---|---|---|---|---|
Level | Estimate | Standard Error | 95% C.I | Z | Pr > |Z| | ||
Intercept | 0.6689 | 0.2510 | 0.1770 | 1.1608 | 2.67 | 0.0077* | |
Sex | Male | − 0.3461 | 0.0778 | − 0.4985 | − 0.1936 | − 4.45 | < 0.0001* |
Wealth index | Middle | 0.0291 | 0.1044 | − 0.1755 | 0.2337 | 0.28 | 0.7805 |
Rich | − 0.3522 | 0.1012 | − 0.5505 | − 0.1540 | − 3.48 | 0.0005* | |
Age of mothers | 20–39 | − 1.0008 | 0.2520 | − 1.4947 | − 0.5068 | − 3.97 | < 0.0001* |
40–49 | − 0.8581 | 0.2670 | − 1.3815 | − 0.3348 | − 3.21 | 0.0013* | |
Antenatal care | 1–4 | − 0.1375 | 0.0986 | − 0.3308 | 0.0557 | − 1.39 | 0.1630 |
Five and above | − 0.2832 | 0.1537 | − 0.5845 | − 0.1181 | − 1.84 | 0.0055* | |
Marital status | Divorced | 0.3402 | 0.1659 | 0.0152 | 0.6653 | 2.05 | 0.0402* |
Widowed | 0.0213 | 0.2419 | − 0.4528 | 0.4955 | 0.09 | 0.9297 | |
Vaccination | Yes | − 0.2582 | 0.1019 | − 0.4580 | − 0.0585 | − 2.53 | 0.0113* |
Anemia | Moderate | 0.2293 | 0.0853 | 0.0622 | 0.3965 | 2.69 | 0.0072* |
Sever | 0.6874 | 0.1887 | 0.3176 | 1.0573 | 3.64 | 0.0003* | |
Education level | Primary | − 0.1962 | 0.1056 | − 0.4031 | − 0.1107 | − 1.86 | 0.0031* |
Secondary and above | − 0.3351 | 0.3350 | − 0.9916 | 0.3215 | − 1.00 | 0.3172 | |
Alpha (\(\alpha\)) QIC = 3986.1527 | 0.4107 | 0.0879 | 0.2385 | 0.5829 | 4.67 | < 0.0001* |
3.2 Comparison of GEE and ALR Models
3.2.1 Parameter Interpretation of Marginal Models
3.3 Analysis of Generalized Linear Mixed Model (GLMM)
Models | AIC | BIC | − 2LogLik | \(\sigma _{W}\) | \(\sigma _{B}\) | p value |
---|---|---|---|---|---|---|
Model with one random intercept | 3933.1 | 4066.8 | 3889.1 | 0.6032 | ||
Model with two random intercept | 3919.7 | 4059.5 | 3873.8 | 0.5392 | 0.2411 | 0.000 |
Effect | Level | Estimates | SE | 95% CI | p value |
---|---|---|---|---|---|
Intercept | 0.7062 | 0.2757 | [0.1657, 1.2468] | 0.0104 | |
Sex | Male | − 0.3815 | 0.0819 | [− 0.5420, − 0.2209] | 0.0079 |
Wealth index | Middle | 0.0650 | 0.1165 | [− 0.1633, 0.2934] | 0.5768 |
Rich | − 0.3304 | 0.1113 | [− 0.5485, − 0.1122] | 0.0029 | |
Age of mothers | 20–39 | − 1.1031 | 0.2502 | [− 1.5937,− 0.6126] | 0.0070 |
40–49 | − 0.9378 | 0.2706 | [− 1.4682, − 0.4074] | 0.0005 | |
Antenatal care | 1–4 | − 0.1557 | 0.1002 | [− 0.3522, 0.0406] | 0.1201 |
\(\ge 5\) | − 0.2956 | 0.1744 | [− 0.6376, − 0.0463] | 0.0002 | |
Marital status | Divorced | 0.3587 | 0.1591 | [− 0.4221, 0.5855] | 0.0242 |
Widowed | 0.0817 | 0.2570 | [0.0467, 0.6707] | 0.7506 | |
Vaccination | Yes | − 0.2640 | 0.1063 | [− 0.4724, − 0.0555] | 0.0130 |
Anemia | Moderate | 0.2459 | 0.0885 | [0.0725, 0.4194] | 0.0054 |
Sever | 0.7822 | 0.1820 | [0.4255, 1.1390] | 0.0016 | |
Education level | Primary | − 0.1577 | 0.1117 | [− 0.3767, − 0.0112] | 0.0030 |
Secondary and above | − 0.4141 | 0.3591 | [− 1.1180, 0.2897] | 0.2488 |