Correspondence 2112 www.thelancet.com Vol 372 December 20/27, 2008 between two interventions, including both treatment and any associated “Hawthorne” effects, since this will best reflect the likely clinical response in practice. 1 In our trial, the control group showed higher than expected rates of improvement and we do consider the possibility of a Hawthorne effect (see Discussion). The other explanation could be natural remission of the disorder. These observations should not, however, divert from the fact that a short period of training and regular supervision of the Lady Health Workers halved the rate of depression in the treated group compared with the control group, highlighting the necessity of strong, well organised health systems that can support such interventions. We were not concerned with strict monitoring of the health workers. The monthly supervision was similar to that which the Lady Health Workers would receive routinely, and this reflected the pragmatic nature of the intervention. Neither was it necessary that all mothers complete the intervention, the analysis being based on intention to treat. This situation is closer to real life where health workers should be able to titrate the intensity and duration of the intervention delivered according to the patient’s response. Unlike in drug trials, it is not possible to do double-blind trials of complex interventions such as “talking” thera- pies: 2 the participants are invariably aware of the type of therapy they receive. However, we followed established procedures for allocation concealment and for ensuring that the researchers measuring outcomes were unaware of the intervention or control status of the clusters. 3 The two groups were carefully randomised, ensuring that the characteristics of patients that might affect outcome, such as immigration or concurrent antidepressant treat- ment, were distributed randomly between treatment and control groups. 4 However, we should clarify that none of the women were on antidepressants at recruitment, and, although they were not stopped from seeking additional treatment, none of the women in the final analysis had been prescribed any antidepressant medication for the duration of the trial. I declare that I have no conflict of interest. Atif Rahman atif.rahman@liverpool.ac.uk University of Liverpool, School of Population, Community, and Behavioural Sciences, Child Mental Health Unit, Alder Hey Children’s NHS Foundation Trust, Mulberry House, Eaton Road, Liverpool L12 2AP, UK 1 Roland M, Torgerson DJ. What are pragmatic trials? BMJ 1998; 316: 285. 2 Day SJ, Altman DG. Blinding in clinical trials and other studies. BMJ 2000; 321: 504. 3 Campbell MK, Elbourne DR, Altman DG. CONSORT statement: extension to cluster randomised trials. BMJ 2004; 328: 702–08. 4 Roberts C, Torgerson D. Randomisation methods in controlled trials. BMJ 1998; 31: 1301. diagnosis of pneumonia could be confirmed. If influenza vaccination leads to better survival of pneumonia, this could mask a true preventive effect. Third, the modifying effect of influ- enza vaccinations in the preceding years was not assessed. In our randomised trial in elderly people, 2 we found that influenza vaccine efficacy was stronger in people who were also vaccinated in the preceding 2 years. Maintenance of seroprotection depended on annually repeated vaccination. 3 Moreover, Jackson and colleagues adjusted for confounding factors defined for the 2 years preceding each season, in- cluding health indicators potentially associated with previous influenza vaccination history. This might have led to overadjustment and, consequently, underestimation of the cumulative preventive effect of repeated vaccinations. In our opinion, the paper by Jackson and colleagues should not be used as a basis to discourage elderly people from receiving an annual influenza vaccination. We declare that we have no conflict of interest. *Carel Thijs, André Knottnerus c.thijs@epid.unimaas.nl School of Public Health and Primary Care, Maastricht University, PO box 616, 6200 MD Maastricht, Netherlands 1 Jackson ML, Nelson JC, Weiss NS, Neuzil KM, Barlow W, Jackson LA. Influenza vaccination and risk of community-acquired pneumonia in immunocompetent elderly people: a population-based, nested case-control study. Lancet 2008; 372: 398–405. 2 Govaert ThME, Thijs C, Masurel N, Sprenger MJW, Dinant GJ, Knottnerus JA. The efficacy of influenza vaccination in elderly individuals: a randomized double-blind placebo-controlled trial. JAMA 1994; 272: 1661–65. 3 Thijs C, Beyer WEP, Govaert ThME, Sprenger MJW, Dinant GJ, Knottnerus A. Mortality benefit of influenza vaccination in elderly people. Lancet Infect Dis 2008; 8: 460–61. Influenza vaccination and risk of community-acquired pneumonia In their case-control study (Aug 2, p 398), 1 Michael Jackson and colleagues found no appreciable asso- ciation between pneumonia and in- fluenza vaccination in elderly people. However, in assessing the effect- iveness of interventions such as this, the case-control approach has serious limitations. First, the high overall proportion of pneumococcal vaccination (95%) among influenza-vaccinated people limited the room for prevention of secondary bacterial pneumonia. The remaining preventive effect of influenza vaccination was further diluted by counting all cases of pneumonia, including those not related to influenza. Second, it is not clear how the investigators dealt with individuals who died, particularly before a Authors’ response Influenza remains an important cause of illness and death in elderly people. We agree with Carel Thijs and André Knottnerus that our Getty Images