Lost to follow-up.

results gained from primary attenders at a tate review clinic after shoulder surgery may not be representative of the outcome: the average self-assessment of the non-attenders was lower by one class. Age differences have been shown to affect the self assessment of health (Idler 1993). In this study, however, with an average age of 5 1.1 years among attenders and 48.8 years in non-attenders, this is an unlikely explanation. Assuming unbiased reporting, the data suggest overestima tion of excellent results and an underestimation of good, fair and poor results. There remains doubt as to whether the non-attenders who were contacted by telephone lack the impact of personal contact in a hospital environment and are therefore less enthusiastic (or more honest) about their outcome. Increas Attempts to study the results of an operation retrospectively are often hampered by the lack of compliance of patients. The outcome among those who fail to attend for follow-up is subject to conjecture. Are they silent because they are dissatisfied (hypothesis A)? Are they so satisfied that they do not want to be bothered (hypothesis B)? Have they died (hypothesis C) or have they just moved house (hypothesis D)? A study on the outcome of shoulder surgery for rupture of the rotator cuff attempted to find an answer. Patients and methods. From 1977 to 1983 a total of 49 patients (51 shoulders) had operations. Of these, 21 patients (43%) accepted an invitation for late review; this included a subjective grading of the outcome. Of the remaining patients, 18 (37%) were traced and contacted by telephone. The average age was similar in both groups (51.1 years among attenders; 48.8 years among non-attenders). The sex ratio in both groups was almost identical (2.5: 1; 2.6:1). Results. Details of the results are shown in Table I. The grading of the primary ‘non-attenders' who were contacted by telephone was worse, supporting hypothesis A against hypothesis B. Three patients had died (hypothesis C). Seven (14%) had moved to an unknown address (hypo thesis D); their outcome was not known and some may have died (hypothesis C). Discussion. The argument that loss of function may be related to the need to change address probably holds true after hip surgery, causing admission to nursing homes. This requires the calculation of a worst-case scenario (Can et at 1993; Murray, Carr and Bulstrode 1993), but the applica tion and/or generalisation of this assumption to patients …