Psychological Medicine

Original Articles

A latent class analysis of drug abuse in a national Swedish sample

K. S. Kendlera1a2a3 c1, H. Ohlssona4, K. Sundquista4a5 and J. Sundquista4a5

a1 Virginia Institute for Psychiatric and Behavioral Genetics, Virginia Commonwealth University, Richmond, VA, USA

a2 Department of Psychiatry, Virginia Commonwealth University, Richmond VA, USA

a3 Department of Human and Molecular Genetics, Virginia Commonwealth University, Richmond, VA, USA

a4 Center for Primary Health Care Research, Lund University, Malmö, Sweden

a5 Stanford Prevention Research Center, Stanford University School of Medicine, Stanford, CA, USA


Background Drug abuse (DA) is a clinically heterogeneous syndrome. Using medical, legal, death and pharmacy records covering the entire population of Sweden, could we uncover meaningful subtypes of DA?

Method We performed a latent class analysis (LCA) on all individuals in Sweden born 1950–1993 who were registered with DA or its consequences (n=192 501) and then validated these classes using demographics, patterns of co-morbidity with alcohol use disorder (AUD), non-DA crime and psychiatric illness, and the pattern of aggregation and co-aggregation in sibling pairs.

Results The best-fit LCA had six classes: (1) low-frequency pure criminal, (2) high-frequency medical criminal, (3) low-frequency pure medical, (4) high-frequency medical, (5) prescription and (6) death. Each class had a distinct pattern of demographic features and co-morbidity and aggregated within sibling pairs with at least moderate specificity. For example, class 2 was characterized by early age at registration, low educational attainment, high male preponderance, high rates of AUDs, strong resemblance within sibling pairs [odds ratio (OR) 12.6] and crime and the highest risk for DA in siblings (20.0%). By contrast, class 5 had a female preponderance, late age at registration, low rates of crime and AUDs, high rates of psychiatric illness, high familiality within sibling pairs (OR 14.7) but the lowest observed risk for DA in siblings (8.9%).

Conclusions DA as assessed by public records is a heterogeneous syndrome. Familial factors contribute substantially to this heterogeneity. Advances in our understanding of etiological processes leading to DA will be aided by a consideration of this heterogeneity.

(Received October 02 2012)

(Revised December 10 2012)

(Accepted January 03 2013)

(Online publication February 01 2013)

Key words

  • Drug abuse;
  • latent class analysis;
  • siblings;
  • Sweden


c1 Address for correspondence: K. S. Kendler, M.D., Virginia Institute for Psychiatric and Behavioral Genetics of VCU, Box 980126, Richmond, VA 23298-0126, USA. (Email: