From the authors:
- Y. Jegal and
- D. S. Kim
- Pulmonary and Critical Care Medicine, Asan Medical Center, Ulsan University, Seoul, Republic of Korea.
- D. S. Kim, Pulmonary and Critical Care Medicine, Asan Medical Center, Ulsan University, Poongnap-dong, Songpa-ku, 138-736, Seoul, Republic of Korea. E-mail: dskim{at}amc.seoul.kr
We thank to H. Chen and co-workers for the interest in our recent paper in the European Respiratory Journal 1, and for raising an important point with good suggestions.
We agree with their opinion, missing data is one of the biggest problems in retrospective studies and we should have discussed this in more detail. However, we still think that the missing data of our study did not markedly influence our results and conclusion 1.
Although there were many missing data in the pulmonary function tests (PFT), the clinical course of most patients were well documented. As stated in our study, 81% of patients had improved or were stable after the initial treatment 1. Although follow-up PFT was variable, as the treatment duration was different for each case, PFT was performed at the time of completion of the initial treatment in all patients, except those who died during the treatment. Death was not the main cause of missing data. Furthermore, the course of many of the missing patients was confirmed by the lung function tests performed several years later, physical check-ups at routine health examination and/or telephone interviews. The reason for loss of follow-up was not because patients were ill, but because they felt too well to visit the hospital. For example, the reason for missing PFT data in 15 patients at 12 months was due to death (n = 7), a worsened condition (n = 1), an improved or stable condition (n = 2), and PFT skipped at 12 months and then performed ≥3.5 yrs after the diagnosis (n = 1). The remaining four patients were lost to follow-up but were confirmed to be clinically stable several years later.
At 2 yrs, there was missing data for 31 patients due to death (n = 11), lost to follow-up in a worse state (n = 1), lost to follow-up in an improved or stable condition (n = 8), and the remaining 11 patients were confirmed to be stable several years later. Of these 11 patients, six did not have available PFT data and were clinically stable and five patients were confirmed to be in an improved or stable condition according to lung function tests performed 3–4 yrs following treatment.
We reanalysed our data according to the recommendation by H. Chen and co-workers. The results of “complete case analysis” and “complete case analysis including missing data” are presented in tables 1⇓ and 2⇓, respectively. In table 2⇓, we put death and cases lost to follow-up in aggravated state in the “worsened” category and patients who were lost during the follow-up but were confirmed to be clinically stable after several years in the “stable” category. Data for cases with some missing PFT data, but who performed PFT later during follow-up, was estimated and added to the appropriate categories. Patients who showed either an improvement or were in a stable condition but were lost completely were put into the “missing” category.
Both tables show that lung function was improved or stable in the majority of the fibrotic nonspecific idiopathic pneumonia patients. An increasing percentage of patients showed improvement of lung function during the first year of follow-up. Although, the number of patients with improvement during the second year seems to be slightly decreased compared to those during the first year, change in forced vital capacity was persistently high in these patients. Table 2⇑ shows that the proportion of real missing patients was not so high to influence the conclusion.
In conclusion, although our data could be partially influenced by “survival bias” in the second year of the treatment and subsequently, the majority of fibrotic nonspecific idiopathic pneumonia patients improved initially and maintained an “improved or stable state” in their lung functions.
ACKNOWLEGEMENTS
We would like to thank to M.S. Lee and J.H. Lee (Ulsan University, Seoul, Republic of Korea) for the consultation regarding statistical analysis.
Statement of interest
None declared.
- © ERS Journals Ltd