Explaining Inconsistencies Between Data on Condom Use and Condom Sales

This article summarises a statistical analysis performed by the authors in their effort to assess the size of the discrepancies between condom sales and estimates of condom usage in developing countries. The authors point out that many HIV prevention programmes use data on condom sales along with survey-based data on condom prevalence to monitor the impact of efforts to change behaviour and improve access. However, these indicators are not always consistent and the paper aims to measure and explain these inconsistencies. In doing so it seeks to assess whether accurate predictions of the number of sex acts and the number of condoms used can be extrapolated from survey data. It is their belief that understanding the apparent inconsistencies between sales and survey data may help clarify the validity of concerns about condom wastage, misreporting, and other related problems.



The authors provide some anecdotal examples of these perceived inconsistencies at the outset, noting that, in Zimbabwe, sales of a socially marketed brand of condoms increased from 1.9 million in 1997, to 4.8 million in 1998, to 8.9 million in 1999, yet nationally representative surveys indicate that condom use in last sex stayed constant between 1996 and 1999 at roughly 34% for males and 17% for females. Some have speculated that these apparent inconsistencies may be attributed to sales figures that are based on sales to the industry rather than actual final sales through to consumers and that there may be relatively large stockpiles sitting in warehouses in many of the countries in question.

The statistical methods employed by the authors in this study are complex and involve several different analysis of data gleaned from several Demographic and Health Studies (DHS) from Kenya, Nigeria, Zimbabwe, and Tanzania between 1994 and 1999. Only one of these surveys (Zimbabwe, 1994) directly asked the participants about their frequency of intercourse, which, along with the percentage of those times in which a condom was used is necessary to estimate the total consumption. As such, the authors used several other techniques such as survival analysis to estimate the frequency of sexual activity and ratio of condom usage from other questions in the survey. These other questions enquired about "time since last intercourse" and "whether a condom was used at last intercourse". The methods involved, and a thorough explanation of the formulas are well discussed in the paper, but are too complex to explain here.

The authors' findings, using these different procedures led to a wide range of estimates of the numbers of condoms used annually. These figures were then compared against the reported sales of condoms (to retailers, not final consumers). The results of the survey analyses, which are based on DHS data, show that the estimates of both the number of sexual acts and the number of condoms used vary enormously based on the estimation method used. For several surveys, the highest estimate of the annual number of condoms used is tenfold that of the lowest estimate. The authors note that while some estimation methods can be disregarded because they yield results that are clearly not plausible, it is impossible to determine which of the remaining methods yield the most accurate results. They also argue that until sales figures to actual consumers are available rather than aggregated amounts to industry members, that the use of these figures for evaluations of condom campaigns effectiveness are of little real value.