Clinical Orthopaedics and Related Research®

, Volume 475, Issue 10, pp 2610–2611 | Cite as

Letter to the Editor: Increased Risk of Revision, Reoperation, and Implant Constraint in TKA After Multiligament Knee Surgery

  • Erfan Ayubi
  • Saeid SafiriEmail author
Letter to the Editor

To the Editor,

In their study, Pancio and colleagues [6], retrospectively evaluated implant survival, complication rate, and long-term outcomes of knee arthroplasty in patients with a prior history of multiligament surgery. The authors found that patients with a prior history of multiligament surgery more frequently received varus-valgus constraint compared with patients undergoing primary knee arthroplasty for osteoarthritis. These results also showed large effect size estimates, but those estimates had wide confidence intervals (CI).

Previous studies [4, 5] showed that wide CIs around large effect-size estimates may indicate sparse-data bias, an estimate inflation that occurs “when the data lack adequate case numbers for some combination of risk factor and outcome levels” [5]. When this occurs, “the resulting estimates of the regression coefficients can have bias away from the null (downward when the estimate is below 1, upward when it is above 1)” [5]. Additionally, sparse-data bias can cause imprecise CIs [5].

Sparse-data bias is generally found if the sample size is relatively small or when there are different sample sizes in the exposure and intervention groups [1, 2, 3, 5, 7]. To address and adjust this bias, researchers developed the conventional method, which adds a half count to every strata in the table. Greenland and colleagues [5] introduced an alternative to the conventional method called penalization via data augmentation, an efficient method “in which external (or prior) information is used to improve accuracy over repeated studies” [5]. Penalization via data augmentation better addresses bias compared to conventional methods by removing or correcting the bias [5].

In Table 4 of the study by Pancio and colleagues, the sample size in the “increased-constraint” group at primary TKA is quite small (n = 9), potentially rendering this dataset to sparse-data bias. We recommend using the penalization method to adjust the sparse-data bias. The results of the penalized regression model showed a decreased effect-size estimate (smaller odds ratios), with narrower and, we believe, more-precise CIs (Table 1).
Table 1

Estimated odds ratios for multiligament group

Method to adjust bias

Estimated odds ratio (95% CI)

Logistic regression

Not applicable

Conventional bias reduction method

44.58 (2.55–780.71)

Penalization via data augmentation method

10.34 (2.18–48.87)


  1. 1.
    Ayubi E, Safiri S. Bias in association between FEV1/FVC % predicted at 7 years and asthma-COPD overlap syndrome. Am J Respir Crit Care Med. 2017;196:115.CrossRefPubMedGoogle Scholar
  2. 2.
    Ayubi E, Safiri S. Lateral lymph node recurrence after total thyroidectomy and central neck dissection in patients with papillary thyroid cancer without clinical evidence of lateral neck metastasis: Comment on data sparsity. Oral Oncol. 2017;69:128.CrossRefPubMedGoogle Scholar
  3. 3.
    Ayubi E, Safiri S. Profile and outcome of first 109 cases of paediatric acute liver failure at a specialized paediatric liver unit in India: Methodological issues. Liver Int. [Published online ahead of print April 9, 2017]. DOI:  10.1111/liv.13455.
  4. 4.
    Greenland S, Mansournia MA. Penalization, bias reduction, and default priors in logistic and related categorical and survival regressions. Stat Med. 2015;34:3133–3143.CrossRefPubMedGoogle Scholar
  5. 5.
    Greenland S, Mansournia MA, Altman DG. Sparse data bias: A problem hiding in plain sight. BMJ. 2016;352:i1981.CrossRefPubMedGoogle Scholar
  6. 6.
    Pancio SI, Sousa PL, Krych AJ, Abdel MP, Levy BA, Dahm DL, Stuart MJ. Increased risk of revision, reoperation, and implant constraint in TKA after multiligament knee surgery. Clin Orthop Relat Res. 2017;475:1618–1626.CrossRefPubMedGoogle Scholar
  7. 7.
    Safiri S, Ayubi E. Dual photon microscopy based quantitation of fibrosis-related parameters (q-FP) to model disease progression in steatohepatitis: Methodological issues. Hepatology. [Published online ahead of print May 16, 2017]. DOI:  10.1002/hep.29261.

Copyright information

© The Association of Bone and Joint Surgeons® 2017

Authors and Affiliations

  1. 1.Department of Epidemiology, School of Public HealthShahid Beheshti University of Medical SciencesTehranIran
  2. 2.Managerial Epidemiology Research Center, Department of Public Health, School of Nursing and MidwiferyMaragheh University of Medical SciencesMaraghehIran
  3. 3.Department of Epidemiology and Biostatistics, School of Public HealthTehran University of Medical SciencesTehranIran

Personalised recommendations