Advertisement

Case-Cohort Studies: Design and Applicability to Hand Surgery

  • Miliana Vojvodic
    Correspondence
    Corresponding author: Miliana Vojvodic, MD, MSc, Division of Plastic and Reconstructive Surgery, Department of Surgery, Faculty of Medicine, University of Toronto, The Rotman/Stewart Building, 149 College St., 5th Floor, Room 508, Toronto, Ontario, M5T 1P5, Canada.
    Affiliations
    Division of Plastic and Reconstructive Surgery, Department of Surgery, University Health Network Division of Plastic and Reconstructive Surgery, Department of Surgery, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada
    Search for articles by this author
  • Mark Shafarenko
    Affiliations
    MD Program, University Health Network Division of Plastic and Reconstructive Surgery, Department of Surgery, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada
    Search for articles by this author
  • Steven J. McCabe
    Affiliations
    Hand Program, University Health Network Division of Plastic and Reconstructive Surgery, Department of Surgery, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada
    Search for articles by this author
Published:April 24, 2018DOI:https://doi.org/10.1016/j.jhsa.2018.03.013
      Observational studies are common research strategies in hand surgery. The case-cohort design offers an efficient and resource-friendly method for risk assessment and outcomes analysis. Case-cohorts remain underrepresented in upper extremity research despite several practical and economic advantages over case-control studies. This report outlines the purpose, utility, and structure of the case-cohort design and offers a sample research question to demonstrate its value to risk estimation for adverse surgical outcomes. The application of well-designed case-cohort studies is advocated in an effort to improve the quality and quantity of observational research evidence in hand and upper extremity surgery.

      Key words

      Within the hierarchy of research designs, observational studies are frequently employed in hand and upper extremity surgery. The practicality and broad application of case-control and cohort studies have established them as important alternatives to their experimental counterparts to query incidence, risk association, and prognosis.
      The case-cohort design is a derivative of the case-control prototype and was initially developed by Prentice in 1986
      • Prentice R.
      A case-cohort design for epidemiologic cohort studies and disease prevention trials.
      to quantify both odds ratios and relative risks. It has garnered increasing popularity for risk estimation among large cohort designs with low incidences of multiple outcomes of interest, where traditional data collection efforts would be laborious and time consuming.
      The defining feature of the case-cohort design is the presence of 2 subject groups within a large patient cohort: a subcohort and a case group.
      • Prentice R.
      A case-cohort design for epidemiologic cohort studies and disease prevention trials.
      • Xue X.
      • Xie X.
      • Gunter M.
      • et al.
      Testing the proportional hazards assumption in case-cohort analysis.
      The case group includes all cohort members that develop an outcome of interest during a period of follow-up after cohort formation.
      • Kulathinal S.
      • Karvanen J.
      • Saarela O.
      • Kuulasmaa K.
      Case-cohort design in practice—experiences from the MORGAM Project.
      This outcome may be defined as an illness or condition, such as a treatment complication, a failure to respond to treatment, or a disease recurrence. The subcohort is a sample of the entire cohort selected at random, without regard to case definition or future outcomes.
      • Sanderson J.
      • Thompson S.G.
      • White I.R.
      • Aspelund T.
      • Pennells L.
      Derivation and assessment of risk prediction models using case-cohort data.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      A subset of individuals may thus become a case within the subcohort
      • Sanderson J.
      • Thompson S.G.
      • White I.R.
      • Aspelund T.
      • Pennells L.
      Derivation and assessment of risk prediction models using case-cohort data.
      (Fig. 1). Importantly, multiple independent case groups can be compared with the common subcohort within a single study.
      Figure thumbnail gr1
      Figure 1Case-cohort study design. The subcohort is created through random sampling of the cohort. As a result, some individuals who become cases are represented in the subcohort.
      Across the roster of observational study designs, the case-cohort offers numerous advantages over related methodologies (Table 1). It is an efficient and resource-friendly method for risk calculation as covariate data is only collected for the subcohort and case groups, rather than the full cohort. This facilitates faster data collection and a corresponding decrease in associated material and human costs.
      • Xue X.
      • Xie X.
      • Gunter M.
      • et al.
      Testing the proportional hazards assumption in case-cohort analysis.
      • Sanderson J.
      • Thompson S.G.
      • White I.R.
      • Aspelund T.
      • Pennells L.
      Derivation and assessment of risk prediction models using case-cohort data.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      • Onland-Moret N.C.
      • van der A.D.L.
      • van der Schouw Y.T.
      • et al.
      Analysis of case-cohort data: a comparison of different methods.
      The random selection of the subcohort also allows for its use as a comparison group for investigating multiple outcomes in 1 study.
      • Xue X.
      • Xie X.
      • Gunter M.
      • et al.
      Testing the proportional hazards assumption in case-cohort analysis.
      • Sanderson J.
      • Thompson S.G.
      • White I.R.
      • Aspelund T.
      • Pennells L.
      Derivation and assessment of risk prediction models using case-cohort data.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      • Onland-Moret N.C.
      • van der A.D.L.
      • van der Schouw Y.T.
      • et al.
      Analysis of case-cohort data: a comparison of different methods.
      • Barlow W.E.
      • Ichikawa L.
      • Rosner D.
      • Izumi S.
      Analysis of case-cohort designs.
      This offers a key advantage over nested case-control designs, in which a new group of controls must be collected for each outcome of interest.
      • Prentice R.
      A case-cohort design for epidemiologic cohort studies and disease prevention trials.
      • Xue X.
      • Xie X.
      • Gunter M.
      • et al.
      Testing the proportional hazards assumption in case-cohort analysis.
      • Kulathinal S.
      • Karvanen J.
      • Saarela O.
      • Kuulasmaa K.
      Case-cohort design in practice—experiences from the MORGAM Project.
      • Sanderson J.
      • Thompson S.G.
      • White I.R.
      • Aspelund T.
      • Pennells L.
      Derivation and assessment of risk prediction models using case-cohort data.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      • Onland-Moret N.C.
      • van der A.D.L.
      • van der Schouw Y.T.
      • et al.
      Analysis of case-cohort data: a comparison of different methods.
      • Barlow W.E.
      • Ichikawa L.
      • Rosner D.
      • Izumi S.
      Analysis of case-cohort designs.
      Table 1Characteristics of Commonly Employed Observational Study Designs
      • Xue X.
      • Xie X.
      • Gunter M.
      • et al.
      Testing the proportional hazards assumption in case-cohort analysis.
      • Kulathinal S.
      • Karvanen J.
      • Saarela O.
      • Kuulasmaa K.
      Case-cohort design in practice—experiences from the MORGAM Project.
      • Sanderson J.
      • Thompson S.G.
      • White I.R.
      • Aspelund T.
      • Pennells L.
      Derivation and assessment of risk prediction models using case-cohort data.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      • Onland-Moret N.C.
      • van der A.D.L.
      • van der Schouw Y.T.
      • et al.
      Analysis of case-cohort data: a comparison of different methods.
      • Barlow W.E.
      • Ichikawa L.
      • Rosner D.
      • Izumi S.
      Analysis of case-cohort designs.
      • Song J.W.
      • Chung K.C.
      Observational studies: cohort and case-control Studies.
      • Ernster V.L.
      Nested case-control studies.
      CohortCase-ControlNested Case-ControlCase-Cohort
      Description
      • Comparison of a study group with risk factors or features of interest with a control group when the outcome of interest occurs
      • Comparison of a case group containing the outcome of interest with a control group without outcome of interest
      • Comparison of cases in the cohort with time-matched control cases
      • Comparison of 1 or multiple case groups with a subcohort control group
      Advantage
      • Exposure data are collected before the disease occurs, allowing for establishment of cause-effect relationships
      • Enables assessment of rare exposures
      • Allows for incidence and relative risk calculation
      • Relatively efficient and inexpensive compared to full cohort studies
      • Enables assessment of rare cases or events as well as outcomes with a long induction or latency period
      • Relatively efficient and inexpensive compared to full cohort studies
      • Risk of recall bias may be lower compared with case-control studies because exposure data are more likely to be collected before disease diagnosis
      • Subcohort allows for lower volume of data collection compared with full cohort
      • Subcohort can be used for comparison with multiple cases
      • Baseline measurement on the subcohort can be taken at any time
      Disadvantage
      • Resource-intensive
      • May be difficult to derive data on rare outcomes
      • May require long follow-up periods for diseases with long latency periods
      • May require large sample sizes in cases of rare outcomes
      • Susceptible to recall or information bias in retrospective studies
      • Risk of recall and information bias if data are collected through self-reporting and interviews
      • Potential difficulty in finding a suitable control group within cohort
      • Requires a defined source population with a predetermined sample size
      • New set of controls need to be identified for each outcome of interest
      • No efficiency benefit for studies with 1 end point/time point
      • Advanced variance computation
      Case-cohort designs also lend themselves to obtaining multiple, timed measurements within the random subcohort and case groups. Baseline measurements on individuals within the subcohort can be taken at any time after the initial cohort formation, whereas nested case-control studies require case identification and definition of controls prior to measurement.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      Finally, case-cohort designs allow for risk ratio estimation even for outcomes that are not rare, in contrast to traditional case-control studies. This also applies to cases in which long follow-up periods are required.
      • Schouten E.G.
      • Dekker J.M.
      • Kok F.J.
      • et al.
      Risk ratio and rate ratio estimation in case-cohort designs: hypertension and cardiovascular mortality.
      • Pearce N.
      What does the odds ratio estimate in a case-control study?.
      Drawbacks of this design relate to complex variance computations when using the asymptotic variance estimator described by Prentice.
      • Prentice R.
      A case-cohort design for epidemiologic cohort studies and disease prevention trials.
      Nevertheless, many common statistical software packages, including SAS and R, allow for the implementation of validated methods of analysis for case-cohorts, including robust variance estimations.
      • Kulathinal S.
      • Karvanen J.
      • Saarela O.
      • Kuulasmaa K.
      Case-cohort design in practice—experiences from the MORGAM Project.

      Guidelines for Reporting

      To reduce methodological heterogeneity and to standardize reporting practice, Sharp et al
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      formulated a comprehensive set of recommendations for studies employing a case-cohort design (Table 2). These recommendations are based on the authors’ own literature review, which highlighted recurring deficiencies in case-cohort utilization and execution.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      As with other observational study methods, the 22-item Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines should be followed to improve transparency and facilitate critical appraisal of any case-cohort study.
      • Vandenbroucke J.P.
      • von Elm E.
      • Altman D.G.
      • et al.
      Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration.
      Table 2Categories of Deficiencies Noted in Existing Case-Cohort Studies and Recommendations for Improvement
      Category of DeficiencyRecommendation
      Study designThe original cohort from which the subcohort was drawn should be described. Definitions of cases, case determination methods, and periods of follow-up should be noted.
      ParticipantsIf exclusion criteria are used, the number of cases, subcohort members, and members of the original cohort should all be provided before and after application of the criteria. Subcohort sampling fraction should also be provided.
      Descriptive informationStudy participants (those included after application of the exclusion criteria) should be described in terms of their exposures and any additional characteristics.
      Statistical methodsThe weighting method used for the Cox proportional hazards model should be stated along with a justification for its use. Kaplan-Meier plots are recommended, in accordance with STROBE guidelines.
      • Vandenbroucke J.P.
      • von Elm E.
      • Altman D.G.
      • et al.
      Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration.
      Adapted from Sharp SJ, Poulaliou M, Thompson SG, White IR, Wood AM. A review of published analyses of case-cohort studies and recommendations for future reporting. PLoS One. 2014;9:e101176.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.

      Applicability to Hand Surgery

      Common barriers to higher-level research in the surgical sciences include funding, time constraints, infrastructure limitations, and ethical concerns.
      • Garas G.
      • Ibrahim A.
      • Ashrafian H.
      • et al.
      Evidence-based surgery: barriers, solutions, and the role of evidence synthesis.
      Observational studies may mitigate some of these impediments by employing a “nonintervention” investigative approach to determine risk associations or treatment effects. In hand research in particular, investigators commonly seek to identify predictors of functional outcome. Outcomes may be determined by a condition’s natural history, an iatrogenic event such as a surgical procedure, or both. In instances in which outcomes of interest are rare, multiple, or of varying timelines, case-cohorts offer an appealing methodology for study. Hand surgeons would invariably benefit from risk assessments for chronic or degenerative conditions of the hand and upper extremity. Further application can also extend to risk factors for adverse outcomes following both surgical and nonsurgical intervention.
      Although the number of case-cohort studies appears to be increasing in recent years, there is still a paucity of literature utilizing case-cohort studies in hand surgery.
      • Sharp S.J.
      • Poulaliou M.
      • Thompson S.G.
      • White I.R.
      • Wood A.M.
      A review of published analyses of case-cohort studies and recommendations for future reporting.
      A search of Medline and EMBASE databases for all case-cohort studies relating to hand surgery published between 2010 and 2017 identified only 2 publications that referenced a case-cohort design.
      • Aleem A.W.
      • Wall L.B.
      • Manske M.C.
      • Calhoun V.
      • Goldfarb C.A.
      The transverse bone in cleft hand: a case cohort analysis of outcome after surgical reconstruction.
      • Ruch D.S.
      • Watters T.S.
      • Wartinbee D.A.
      • Richard M.J.
      • Leversedge F.J.
      • Mithani S.K.
      Anatomic findings and complications after surgical treatment of chronic, partial distal biceps tendon tears: a case cohort comparison study.
      Neither study has utilized a classic case-cohort design, however. Here, we present a sample research question for the initiation and completion of a case-cohort study.

      Sample Experimental Design

      Cubital tunnel syndrome is the second most common compression neuropathy in the upper extremity.
      • Assmus H.
      • Antoniadis G.
      • Bischoff C.
      • et al.
      Cubital tunnel syndrome—a review and management guidelines.
      Simple surgical release of the ulnar nerve is a well-established treatment option for symptomatic cubital tunnel syndrome.
      • Tapadia M.
      • Mozaffar T.
      • Gupta R.
      Compressive neuropathies of the upper extremity: update on pathophysiology, classification, and electrodiagnostic findings.
      • Dellon A.L.
      Review of treatment results for ulnar nerve entrapment at the elbow.
      Multiple complications and surgical failures requiring revision surgery have been described following simple ulnar nerve release. For illustrative purposes, we limit our discussion of potential complications to persistent nerve compression symptoms, paresthesia around the operative site, and injury to the ulnar nerve.
      • Boone S.
      • Gelberman R.H.
      • Calfee R.P.
      The management of cubital tunnel syndrome.
      • Jariwala A.
      • Bansal N.
      • Nicol G.
      • Shelton J.
      • Wigderowitz C.A.
      Outcome analysis of cubital tunnel decompression.
      The patient and surgical factors that are associated with an increased risk for each type of adverse postsurgical outcome are currently unknown. Furthermore, each complication may occur at a different time point following surgery. Early complications include scar tenderness, whereas failure of symptom resolution may not be defined for up to 2 years following the procedure.
      The variety of potential causes and timelines preceding surgical failure makes clinical research in this area time-involved and challenging. A case-cohort design could serve to delineate which factors are antecedent to postdecompression complication or failure. Table 3 and Figure 2 describe a sample research proposal and a possible study algorithm, respectively. Unlike a nested case-control design, a case-cohort would allow for efficient assessment of multiple types of complications (cases), by comparing 1 control subcohort with each case group at multiple time points, if required. Collectively, these results would allow for an independent risk profile for each adverse outcome and could improve presurgical patient selection and informed discussion of risks for simple decompression of the ulnar nerve. Other common pathologies in hand surgery such as basal joint arthritis and distal radius fractures would similarly benefit from application of a case-cohort design to query risk factors associated with poor surgical outcomes.
      Table 3Sample Research Proposal for Case-Cohort Analysis
      Research questionFor symptomatic cubital tunnel syndrome, what risk factors are associated with surgical failure following simple release of the ulnar nerve at the elbow?

      Surgical failures of interest are persistent nerve symptoms in the hand, nerve symptoms around the elbow, and injury to the ulnar nerve.
      CohortLarge number of patients across a large, collaborative multicenter group with symptomatic cubital tunnel syndrome treated with surgical decompression.
      SubcohortRandomly selected subpopulation from original cohort
      Hypothesized risk factors
      • Age
      • Gender
      • Worker’s compensation background
      • Comorbidity burden
      • Previous trauma to the affected upper extremity
      • Smoking history
      • Clinical disease severity (eg, interossei weakness, hypothenar wasting)
      • Nerve conduction velocities
      • Clinical disease duration
      • Prior subluxation of ulnar nerve
      • Placement of incision at the elbow
      Cases
      • 1.
        Persistent ulnar nerve compression symptoms in the hand
      • 2.
        Nerve symptoms at the level of the elbow
      • 3.
        Injury to the ulnar nerve
      Figure thumbnail gr2
      Figure 2Schematic presentation of case-cohort design with sample patient numbers (n). Three different outcomes of failure (ulnar nerve symptoms in the hand, ulnar nerve symptoms at the elbow, injury to the ulnar nerve) are associated with risk factors through a randomly sampled subcohort group. The same subcohort group can be used for comparison for all 3 outcomes of interest.

      Summary

      The case-cohort design offers numerous advantages over alternate methodologies for risk estimation. Guidelines for reporting of case-cohort studies have been published to improve their validity and uniformity. The utilization of case-cohort studies in hand and upper extremity research is limited. The practice of hand surgery would benefit from the use of case-cohort study designs to determine risk factors for disease development or adverse outcomes.

      References

        • Prentice R.
        A case-cohort design for epidemiologic cohort studies and disease prevention trials.
        Biometrika. 1986; 73: 1-11
        • Xue X.
        • Xie X.
        • Gunter M.
        • et al.
        Testing the proportional hazards assumption in case-cohort analysis.
        BMC Med Res Methodol. 2013; 13: 88
        • Kulathinal S.
        • Karvanen J.
        • Saarela O.
        • Kuulasmaa K.
        Case-cohort design in practice—experiences from the MORGAM Project.
        Epidemiol Perspect Innov. 2007; 4: 15
        • Sanderson J.
        • Thompson S.G.
        • White I.R.
        • Aspelund T.
        • Pennells L.
        Derivation and assessment of risk prediction models using case-cohort data.
        BMC Med Res Methodol. 2013; 13: 113
        • Sharp S.J.
        • Poulaliou M.
        • Thompson S.G.
        • White I.R.
        • Wood A.M.
        A review of published analyses of case-cohort studies and recommendations for future reporting.
        PLoS One. 2014; 9: e101176
        • Onland-Moret N.C.
        • van der A.D.L.
        • van der Schouw Y.T.
        • et al.
        Analysis of case-cohort data: a comparison of different methods.
        J Clin Epidemiol. 2007; 60: 350-355
        • Barlow W.E.
        • Ichikawa L.
        • Rosner D.
        • Izumi S.
        Analysis of case-cohort designs.
        J Clin Epidemiol. 1999; 52: 1165-1172
        • Schouten E.G.
        • Dekker J.M.
        • Kok F.J.
        • et al.
        Risk ratio and rate ratio estimation in case-cohort designs: hypertension and cardiovascular mortality.
        Stat Med. 1993; 12: 1733-1745
        • Pearce N.
        What does the odds ratio estimate in a case-control study?.
        Int J Epidemiol. 1993; 22: 1189-1192
        • Vandenbroucke J.P.
        • von Elm E.
        • Altman D.G.
        • et al.
        Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration.
        PLoS Med. 2007; 4: e297
        • Garas G.
        • Ibrahim A.
        • Ashrafian H.
        • et al.
        Evidence-based surgery: barriers, solutions, and the role of evidence synthesis.
        World J Surg. 2012; 36: 1723-1731
        • Aleem A.W.
        • Wall L.B.
        • Manske M.C.
        • Calhoun V.
        • Goldfarb C.A.
        The transverse bone in cleft hand: a case cohort analysis of outcome after surgical reconstruction.
        J Hand Surg Am. 2014; 39: 226-236
        • Ruch D.S.
        • Watters T.S.
        • Wartinbee D.A.
        • Richard M.J.
        • Leversedge F.J.
        • Mithani S.K.
        Anatomic findings and complications after surgical treatment of chronic, partial distal biceps tendon tears: a case cohort comparison study.
        J Hand Surg Am. 2014; 39: 1572-1577
        • Assmus H.
        • Antoniadis G.
        • Bischoff C.
        • et al.
        Cubital tunnel syndrome—a review and management guidelines.
        Cent Eur Neurosurg. 2011; 72: 90-98
        • Tapadia M.
        • Mozaffar T.
        • Gupta R.
        Compressive neuropathies of the upper extremity: update on pathophysiology, classification, and electrodiagnostic findings.
        J Hand Surg Am. 2010; 35: 668-677
        • Dellon A.L.
        Review of treatment results for ulnar nerve entrapment at the elbow.
        J Hand Surg Am. 1989; 14: 688-700
        • Boone S.
        • Gelberman R.H.
        • Calfee R.P.
        The management of cubital tunnel syndrome.
        J Hand Surg Am. 2015; 40: 1897-1904
        • Jariwala A.
        • Bansal N.
        • Nicol G.
        • Shelton J.
        • Wigderowitz C.A.
        Outcome analysis of cubital tunnel decompression.
        Scott Med J. 2015; 60: 136-140
        • Song J.W.
        • Chung K.C.
        Observational studies: cohort and case-control Studies.
        Plast Reconstr Surg. 2010; 126: 2234-2242
        • Ernster V.L.
        Nested case-control studies.
        Prev Med. 1994; 23: 587-590