Agreement between head of household informant and self-report in a community survey of substance use in India

Abstract

This survey of 500 households in a New Delhi urban slum compared reports of substance use by the head of the household informant with individual self-report. Information from the two sources was compared for 1132 people above the age of 15 years. The paired agreement regarding the use of substances was high (κ=0.92; S.E.=0.01, z=92.0). The agreement regarding the presence of symptoms and classification of dependence for the use of alcohol, tobacco and opiates ranged from good to excellent and head of household reports had a high positive predictive value for the use of these substances. This method provides useful estimates of drug use and dependence for substances associated with observable physiologic withdrawal syndromes, and is less costly and quicker to perform than traditional self-report methodologies.

Introduction

Information on substance use patterns in the population is generally derived from community surveys (Johnston, 1980, Johnston, 1989) using self-reports obtained through face-to-face interviews (Grant, 1997, Caetano, 1999). Like all assessment methods, self-report surveys have their limitations. Time and money are consumed in trying to locate the entire sample designated for interview in a given survey. Underestimates of actual substance use and dependence can result from the inability to locate and directly interview the more severely affected users due to their increased mobility and itinerancy. Self-reported data can also result in under-reporting of substance use (McAllister and Makkai, 1991, Embree and Whitehead, 1993, Romelsjo et al., 1995).

Informant methodologies have been proposed as a useful adjunct to self-report survey techniques (Smart and Liban, 1982). Proxy informants have been used to gather information in a variety of settings. Dietary information has been collected from proxy informants and used in the study of risk factors for development of various diseases (Moore et al., 1970, Kolonel et al., 1977, Marshall et al., 1980, Humble et al., 1984). The informant method has also been used to assess consumption of alcohol in a Mexican community (Natera et al., 1985) and in an urban setting in New Zealand (Graham and Jackson, 1993). Information obtained through informants in both these studies agreed closely with the primary respondent, especially with regard to the frequency of use of substances. Information from a proxy informant has also been employed in gathering information about the physical health status and disability levels in the urban community setting in the United States (Magaziner et al., 1996). Informant reports regarding the performance of various activities of daily living agreed almost perfectly with individual reports. Agreement was also very good regarding the identification of five out of nine chronic conditions in the individual. The proxy respondents were found to produce reports that agreed much more precisely if they lived with, or spent a great deal of time with the primary respondent.

The cohesiveness of the family in India is still largely intact. The Indian society consists of closely-knit families in both metropolitan and rural areas that generally reside together in joint settings along paternal lines. Due to close proximity among family members there is little privacy and the head of the family is generally aware of the behaviour and activities of the other family members. A study of drug use in Nepal, a country with similar social structure, suggests that the even the vast majority of illicit drug users continue to reside at home with their families in joint settings (Chatterjee et al., 1996). A report regarding the profile of illicit drug use in New Delhi reports that the majority of illicit drug users are married and live with their families of the type described by us in the community (Ray et al., 2000).

Countries such as India have great monetary constraints limiting all health care research activities. Less costly techniques that give valid and reliable results in survey research of substance use would be helpful in boosting surveillance activities which are onerous in such a vast and diverse population. It was postulated that the head of household informant has enough contact with other members of the household to enable correct identification of those individuals who are using drugs. Furthermore, the household informant is able to describe patterns of use and observable signs and symptoms of dependence. A pilot study was conducted to test the agreement of information provided by the head of the household and the individual on self-report in an urban slum in New Delhi, India (Mohan et al., 1992). The present communication carries the earlier report further by reporting on the analysis on agreement, sensitivity, specificity and predictive value of information provided by the household informant.