Socioeconomic inequalities in drug utilization for Sweden: Evidence from linked survey and register data Martin Nordin a, b, * , Margareta Dackehag a, b , Ulf-G. Gerdtham a, b, c a Department of Economics, Lund University, P.O. Box 7082, 220 07 Lund, Sweden b Health Economics & Management, Institute of Economic Research, Lund University, P.O. Box 7080, 220 07 Lund, Sweden c Centre for Primary Health Care Research, Lund University, Sweden article info Article history: Available online 22 November 2012 Keywords: Sweden Health inequality Healthcare Drug utilization Income Education abstract This study analyzes the socioeconomic gradient in drug utilization. We use The Swedish Prescribed Drug Register, merged with the Survey of Living Conditions (the ULF), and the study sample consists of 8138 individuals. We find a positive education gradient (but no income gradient) in drug utilization, after controlling for health indicators. Whereas high-educated men use a larger number of drugs, high-educated women use both a larger number of drugs and more expensive drugs. For males, but not as clearly for females, we find that the education gradient is weaker for more health-related drugs but stronger for more expensive drugs. We conclude that the main reason for the education gradient in drug utilization is doctors’ behaviour rather than compliance with medication and affordability of drugs. Ó 2012 Elsevier Ltd. All rights reserved. Introduction There are large disparities in health by socioeconomic status (SES) in Sweden and throughout the world (see for example; Nordin & Gerdtham, 2010; Baum & Ruhm, 2009; Buckley, Denton, Robb, & Spencer, 2004; Deaton & Paxton, 1998; Gerdtham & Johannesson, 2000, 2002, 2004; Smith, 2004; van Doorslaer et al., 1997; Wagstaff & van Doorslaer, 2000; van Doorslaer & Koolman, 2004). Although the link between health and SES is not fully understood, differences in access to adequate healthcare might partly cause SES- health disparities (van Doorslaer, Koolman, & Jones, 2004; van Doorslaer et al., 2006; Gerdtham & Trivedi, 2001; Morris, Sutton, & Gravelle, 2005; Sutton et al., 2002). Studies show, for example, that there is SES inequality in physician utilization and inpatient care in Sweden and other countries, and that the affluent account for the larger number of specialist visits in most OECD countries (Sutton et al., 2002; van Doorslaer et al., 2004). Another healthcare aspect that might cause the SES-health gradient is differences in the availability and utilization of (medical) drugs. Non-compliance with medication is a major health-related problem (Larsen, Stovring, Kragstrup, & Hansen, 2009), and for most diseases only 50 percent of the patients seems to take their prescription drugs correctly (Morris & Schultz, 1992; WHO, 2003). Against this background, the objective of this study is to find out whether there is a positive association between SES and drug utilization in Sweden, i.e. if there is SES inequality in drug utilization. The main explanation for a SES gradient in drug utilization is, of course, that health is the main predictor of drug utilization and health is (as acknowledged above) related to SES. In fact, such a negative correlation between drug utilization and SES has been established for Sweden (Henricson et al., 1998; Weitoft, Rosén, Ericsson, & Ljung, 2008). Therefore, it is fundamental to control for health, otherwise we would again just be estimating the familiar SES-health gradient when using drug utilization as a proxy for health. In Sweden studies have found positive associations between women’s education level and hormone replacement therapy (Merlo et al., 2003), and between education level and dementia drugs (Johnell, Weitoft, & Fastborn, 2008). However, medical liter- ature on the relationship between SES and drug utilization shows no consistency in results, possibly because of small and specific subpopulations (Nielsen, Hansen, & Rasmussen, 2003). Another drawback of these studies is that they often use self-reported drug utilization (for example Furu, Straume, & Thelle, 1997; Nielsen et al., 2003), a measure that is likely to be plagued with measurement errors. Illness or disabilities might for example seriously influence the process of recalling information. We use The Swedish Prescribed Drug Register, merged with the Survey of Living Conditions (the ULF), and register data from * Corresponding author. Department of Economics, Lund University, P.O. Box 7082, 220 07 Lund, Sweden. Tel.: þ46 46 222 46 69. E-mail address: Martin.Nordin@nek.lu.se (M. Nordin). Contents lists available at SciVerse ScienceDirect Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed 0277-9536/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.socscimed.2012.11.013 Social Science & Medicine 77 (2013) 106e117