NEURO-EPIDEMIOLOGY Classification of surgical procedures for epidemiologic assessment of sporadic Creutzfeldt-Jakob Disease transmission by surgery J. de Pedro-Cuesta 1 , M. J. Bleda 1 , A. Ra´bano 2 , M. Cruz 3 , H. Laursen 4 , K. Mølbak 5 & A. Siden 3 on behalf of the EUROSURGYCJD Research Group 1 Department of Applied Epidemiology, National Center for Epidemiology, Carlos III Institute of Health, C/Sinesio Delgado 6, 28029, Madrid, Spain; 2 Neuropathology Laboratory, Alcorco´n Hospital, Avda Budapest 1, Alcorcon, 28922, Spain; 3 Department of Clinical Neurosciences, Neurology Division, Karolinska Institutet, SE-141 86, Stockholm, Sweden; 4 Neuro- pathology Laboratory, 6301. H:S Rigshospitalet, Blegdamsvej, 9, DK-2100, Copenhagen, Denmark; 5 Department of Epide- miology, Statens Serum Institut, Artillerivej 5, DK-2300, Copenhagen, Denmark Accepted in revised form 17 July 2006 Abstract. Background: In this preparatory phase of a case–control study, we propose and evaluate a new tool for classifying surgical procedures (SPs) in cat- egories useful for epidemiologic research on surgical transmission of sporadic Creutzfeldt-Jakob disease (sCJD). Methods: All SPs reported to the Swedish National Hospital Discharge Registry in the period 1974–2002, and undergone by 212 Swedish patients with registered diagnosis of CJD at death, hospital discharge or notification, in the period 1987–2002, 1060 age-, sex- and residence-matched controls and 1340 randomly chosen population controls, were reclassified into one of six categories of hypothetical transmission risk level. For that purpose the follow- ing two attributes were used: non-disposable instru- ments involved; and highest assigned ad-hoc risk level for four tissues or anatomical structures contacting such instruments. Results: A total of 1170 different SP codes were reclassified as follows: 3.1% in the high-risk, 59.1% in the lower-risk, 24.4% in the low- est-risk, and 2.1% in the no-risk groups, with 11.3% procedures negatively defined by rubric as ‘‘other than...’’ being assigned to two spurious diluted-high and diluted-lower risk categories. The high-risk group mainly comprised neurosurgical (53%) and ophthalmic (39%) procedures. Sensitivity of neuro- surgery and of ophthalmic surgery excluding neuro- surgery, for the high- and diluted-high risk vs. other categories was 46% and 84%, while specificity was 98% and 95%, respectively. Sensitivity analysis based on these indices revealed that non-significant odds ratio effects of 1.4 and 1.3 for neurosurgery and ophthalmic surgery corresponded to statistically sig- nificant values of 5.1 after reclassification. Conclu- sions: This classification might contribute to quantify effects masked by use of body-system SP-categories in case–control studies on sCJD transmission by sur- gery. Key words: Care, Creutzfeldt-Jakob Disease, Epidemiology, Etiology, Methods, Safety, Surgical procedures Introduction The need to study the association between surgery and human transmissible spongiform encephalopa- thies (HTSE) has been highlighted [1–3]. Surgical transmission of sporadic CJD (sCJD) has been ad- dressed by six case–control studies, in many instances with diverging and partly inconsistent results [4–10] and one meta-analysis [11]. In terms of design, the above-mentioned studies might present common methodologic limitations due to possible case-selec- tion or -recall bias, or more specific limitations, such as asymmetry through use of clinical controls [4–7], surrogate informants solely for cases [4, 5, 8, 10], and different calendar time intervals for operations on cases and controls [8, 10]. In addition, the use of a surgical procedure (SP) classification insufficiently specific or sensitive to distinguish between low- and high-risk interventions for sCJD transmission, could constitute a source of misclassification bias. For example, a recent British report found that multiple categories of ophthalmic- and neurosurgical proce- dures, thought to be at highest risk for surgical instrument contamination, were in fact not liable to encounter potentially infective tissues [12]. A reported-instrument-based [13] classification of tissues and anatomical structures was recently developed by our group, taking into account contact with surgical instruments carrying tissue remnants potentially relevant for person-to-person transmis- sion of sCJD [14]. However, an operational catego- rization of single, well-defined SPs in terms of European Journal of Epidemiology (2006) 21:595–604 Ó Springer 2006 DOI 10.1007/s10654-006-9044-7