Indmedica Home | About Indmedica | Medical Jobs | Advertise On Indmedica
Search Indmedica Web
Indmedica - India's premier medical portal

Indian Journal of Community Medicine

Analysis of Adoption Preference of Family Planning Methods Through Multiple Logistic Regression

Author(s): A.R. Aggarwal, P. Singh

Vol. 29, No. 2 (2004-04 - 2004-06)

Summary

The present paper attempts to identify the motivating factors for adoption of family planning practices through logistic regression analysis. The data for this analysis relates to about 1 160 eligible couples from Paharganj area of Delhi of which 556 are non users, 176 are IUD acceptors and 428 are sterilization acceptors. The multiple logistic regression has been applied for IUD acceptors vs non users, sterilization vs non users and IUD vs sterilization. It has been observed that female literacy is a significant motivating factor for the adoption of family planning practices. Hence this suggests that it was female literacy which influenced the adoption of family planning methods.

Key words: Logistic regression, motivating factors.

Introduction

Multiple Logistic Regression Analysis is a powerful multivariate technique which has been applied to the situation where the data is in binary form i.e., in the form of O and 1. This technique is extensively used in medical research in identifying the risk factors of a disease. If, the independent variables are also in the binary form, the coefficients of the multiple logistic model have nice interpretation, as exponential of these represent the relative risks in developing the disease. The usefulness of logistic regression model has been illustrated in identifying the preference factors for the adoption of spacing and terminal methods of family planning.

Material and Methods

The area of Paharganj in New Delhi has been allotted to Lady Hardinge Medical College (LHMC) Delhi by Delhi administration for promoting family planning acceptance by providing both educational and motivational programmes as well as services. This area has about 10,000 households. The Institute for Research in Medical Statistics-has collected the data on the adoption of family planning practices in this area during 1987-1988 in collaboration with LHMC & Smt. Sucheta Kripalani Hospital. This data has been used to show the application of logistic regression, a multivariate technique in family planning. The data has been collected on 2,349 eligible couple, of whom 1,535 were family planning users and 814 were non acceptors. The analysis is based on 1,160 eligible couple for which the complete information is available. Among 1,160 couples 556 are non users, 176 are IUD acceptors and 428 are sterilization acceptors. The multiple logistic regression analysis has been applied for identifying the motivating factors for the adoption of family planning methods.

Logistic Model

The logistic model specifies that the probability of disease depends on set of variables x l, x2 xp in the following way px=p(y=1/x)

= 1I (1 + exp {-30 + 31 x 1+32x2+ ....... +Bpxp)

where y denotes either the presence ( y =1) or absence (y = 0) of disease and x denotes a set of p variables which may represent the risk factors. The 3,s are the parameters that represent the effect of x's on the risk of disease.

In terms of logic

log (px/gx) = B0+B1x1+B2x2+ ......... B pxp

This is similar to multiple linear regression. The parameters B i's are called logistic regression coefficients and are estimated by maximum likelihood method. The exponent of these 3s is known as odds ratio which estimates the relative risk in case control studies, if X is in binary form. In the present case it will be referred to as adoption preference for family planning methods.

The logistic Regression was applied to study the following situations: (i) IUD users versus non-users, (ii) Sterilization acceptors versus non users, (iii) Sterilization acceptors versus IUD users.

The three independent variables or motivated variables considered are education of wife, number of male children and number of female children. Separate coding has been done for these variables. Female literacy has been given three types of codes viz, illiterates, literate up to higher secondary and literate above higher secondary level. Dummy variables have been introduced for this purpose. All illiterate females are coded as (0,0) up to higher secondary as (1,0) and above higher secondary as (0,1). For number of male children, couples having up to 1 male child have been coded as 0 and those having 2 and more male children as 1. Similar coding has been done for number of female children. The analysis has been done by a computer programme MULTLR. This program is written by McGee, DL.

Results and Discussion

The results for multiple logistic regression for three situations considered are presented in Tables I-III. The relative motivation of family planning is estimated by comparing the illiterates. The motivation variables are education of wife, literacy up to higher secondary and above higher secondary. For number of male children couples having two or more male children has been compared with couple having only one male child. Similar categories have been made for number of female children.

Table I : Multiple Logistic Regression : IUD user versus Non user

VARIABLE COEFF S.E. Z.SCORE P. VALUE OR. 95 % CI
X 1 1.42 0.27 5.23 <0.0001 4.12 2.42-7.01
X2 2.14 0.28 7.67 <0.0001 8.53 4.93-14.74
X3 1.21 0.19 6.44 <0.0001 3.36 2.32-4.86
X4 0.24 0.20 1.21 0.23 1.27 0.86-1.87
Constant -3.63 0.28 -13.13 <0.0001    

Table II : Multiple Logistic Regression Sterilization Acceptor versus Non user

VARIABLE COEFF S.E. Z.SCORE P. VALUE OR. 95 % CI
X1 0.38 0.16 2.41 0.0160 1.46 1.07-1.98
X2 0.54 0.19 2.80 0.0052 1.71 1.17-2.48
X3 2.36 0.15 15.46 <0.0001 10.6 17.86-14.32
X4 0.78 0.14 5.588 <0.0001 2.19 1.66-2.88
Constant -2.726 0.178 -15.309 0.0000    
X1 x X2 = female literacy X3 = number of male children, X4 = number of female children

Table III : Multiple Logistic Regression IUD user versus Sterilization Acceptor

VARIABLE COEFF S.E. - Z.SCORE P VALUE OR. 95 % CI
X 1 -0.95 0.29 -3.30 0.0010 0.38 0.22-0.68
X2 -1.53 0.30 -5.17 <0.0001 0.22 0.12-0.39
X3 1.62 0.23 7.09 <0.0001 5.05 3.22-7.89
X4 1.21 0.23 5.25 <0.0001 3.35 2.13-5.26
Constant 0.26 0.32 0.80 0.43

TABLE IV: Adoption Preferences and Measures

CTOR IUD
v/s Non users
Sterilization
v/s IUD users
Sterilization
v/s non users
Female literacy
(a) Up to HS
v/s illiterates
4.2 1.4 0.39
(b) Higher education vs illiterates 8.5 1.7 0.22
Number of children
2 male children **
vs up to 1 male child
3.3 10.3 5.04
2 female children *** 1.2 2.19 3.35

Reference category - illiterate, ** Reference category < 2 male children, *** Reference category has been 2 female children.

In table IV the figures are the relative preferences (estimated as exponent of regression coefficient) for an eligible couple to accept F.P method. In the column corresponding to acceptors of IUD users versus non user of F.P method female literacy plays an important role. If the literacy of female was more than H.S. then an eligible couple had 8.5 times preference for accepting IUD method as compared to couples with illiterate females. Similarly' if a female is educated up to HS level then she had 4.2 times preference for IUD use as compared to illiterate female. Number of male children also was a motivating factor for the use of IUD method of F.P. A couple with 2 or more male children has 3.3 times motivation or preference for accepting IUD as compared to those having up to 1 male child.

For sterilization versus non acceptors, the motivating force appeared to be mainly the number of male children. A couple with 2 or more male children had 10.6 times preference I motivation for undergoing the sterilization as against those having up to 1 male child. Number of female children also motivates a couple to undergo sterilization. Although female literacy had significant impact, the motivating factor was only 1.7 far higher education and 1.4 up to HS education.

As to the preference for sterilization versus IUD use, the results in the last column indicate that number of children male as well as female determine the relative preference for a couple in favour of sterilization as compared to IUD use, the factor being 5.0 for number of male children and 3.3 for number of female children. The level of female literacy also determine the relative preference for IUD use as compared to sterilization, it being 2.6 for up to H.S. education and 4.5 for higher level of education versus illiterates.

Summary

The logistic regression model has been applied to family planning data to estimate the preference factors for the adoption of spacing and terminal methods of family planning with respect to female literacy and number of children. The main motivating factor for adopting spacing method was observed as female literacy and for adopting terminal method was the number of male children. The relative preference for these two factors were observed as 8.5 and 10.6 respectively.

References

  1. Anderson JA Diagnosis by logistic discriminant function : further practical problems and results: Journal of the Royal Statistical Society, 1974; C 23 : 397-404.
  2. Mc Gee DL A program for logistic regression on the IBM PC. Amer. J. Epidemiol 1986; 124: 702-705.
  3. Walter SD, Feinstein AR, and Wells CK Coding ordinal independent variables in multiple regression analysis. Amer. J. Epidemiol 1987; 125: 319-323.
  4. Schlesselman JJ Case control studies : Design Conduct, Analysis. Oxford University Press, New York.
  5. Anderson JA (1973) Logistic discrimination with medical application. pp 1-15 Academic Press, New York
  6. Anderson JA (1991) et al Analysis of adoption preference of Family Planning Acceptors through Discriminant Factors. The Journal of Family Welfare, 1991; xxxvii : 13-18.

Institute For Research in Medical Statistics (ICMR), Ansari Nagar, New Delhi -110029

Access free medical resources from Wiley-Blackwell now!

About Indmedica - Conditions of Usage - Advertise On Indmedica - Contact Us

Copyright © 2005 Indmedica