Circulation 60,475 • Volume 15, No. 4 • Winter 2000

Postoperative Data Collection and Outcome

Terri Monk, MD

The first step in any research effort is identification of the specific problem that will be investigated. This step is critical because the research question controls the direction of the subsequent study design, data collection, and analysis. When identifying the research question, three general criteria must be considered to determine if it is worth pursuing1:

Is the question important? Practice parameters from the American Medical Association2 define outcome research as “the study of the results of medical interventions.” In clinical medicine, good research questions are usually patient oriented. In outcome research, these questions often address patient morbidity or mortality, quality of life, or economic issues. These questions should pass the “so what” test; getting the answer should contribute substantially to medical knowledge.3 Is the question answerable? Is it feasible?

Once the research question is defined, the researcher must determine what data to collect and how the data will be collected. There are two types of data collectionÑpassive or active.4 In passive research, the researcher waits for the problem to develop and then investigates the cause. The most common type of study design for passive research is the case-control study. In active research, data is collected before the problem develops and the researcher waits for the outcomes to become evident. Active research is usually done with a prospective cohort study or a randomized clinical trial.

At the present time, I am doing outcome research evaluating postoperative cognitive decline (POCD). If I were using a case-control study design to evaluate this outcome, I would wait for patients to develop this problem and then identify several patients who had the same type of surgery but did not develop POCD. These non-affected patients are the controls and they are usually matched to the patient by age and other demographic characteristics. Case-control studies are generally performed because they are less expensive than prospective cohort studies. However, they are retrospective and do not establish a sequence of events. Another major limitation of case-control studies is that they are susceptible to bias. The bias comes from two sources: the separate sampling of the cases and controls and the retrospective measurement of the predictor variable.3 To avoid these problems, I am using a prospective cohort study design in my research. With this study design, I identify patients before their surgery and follow these patients after surgery to see if they develop POCD. Following these patients over time allows the researcher to identify the incidence of the problem and collect data on patient factors. The prospective cohort design is especially good for studying the antecedents of a serious problem and the only way to establish the temporal sequence of predictors and outcome variables.3 The main disadvantages of this study design are that it requires large number of patients, especially for rare events, and is much more expensive than a case-control study. The third type of study design is the randomized, clinical trial. This experimental design is generally considered to be the gold standard for clinical research. In a randomized, clinical trial, the investigator manipulates the intervention and observes the effects on an outcome.3 This type of study design is not possible for outcome research in anesthesiology at the present time because too little is known about the incidence and potential etiologies of the outcome problems.

Where does one start in the investigation of outcome problems in Anesthesiology? The Institute of Medicine report entitled “To Err is Human” strongly recommends the development of both mandatory and voluntary error reporting systems.5 Mandatory reporting would focus on the identification of serious adverse events, including death and serious injuries, attributable to error. A voluntary reporting system of errors that result in minimal or no patient harm is also recommended. It seems obvious that some type of error reporting system will be developed for all specialties of medicine in the near future. I would like to see the Anesthesia Patient Safety Foundation lead in the development of such an error reporting system for our specialty rather than waiting for the federal government to mandate the system.

Once an outcome research plan is identified, the next step is the development of data collection techniques. At the present time, data collection for clinical research is often handwritten and then keyed into a computer, leading to great potential for data errors. Manual review of the data entry is required to determine that valid and complete data has been entered on the forms. Scanable coding forms can eliminate keying errors, but are still labor intensive, requiring a person to both fill out the forms and scan the data into the computer. Computer-aided input at the site eliminates errors in data entry, but still requires complex procedures to check data against previously entered data and to ensure that all the necessary data is entered.

The new paradigm in data collection is web-based data collection.6 As soon as a subject enters the study, a message is electronically forwarded to a data-coordinating center. Case report forms are than transmitted via the web to the study site. Once the form is filled out, it can be returned to the data-coordinating center, thus eliminating the need for storage of forms at the study site. The only hardware requirement at the research site is a standard computer and an appropriate internet browser. Web-based data collection significantly reduces the administrative effort at the local site. Site training for study participation can be done online and there can be automatic documentation by the data-coordinating center that successful training has been completed. This simplified research process should allow private practice physicians to participate in data collection because there will be minimal increased management burdens and space requirements for the conduct of research.

If any national outcome reporting system is to be successful, confidentiality of data must be guaranteed. Web-based research simplifies and improves security issues. All access to the study is through a central web that can encrypt all transmissions and restrict individual access to data. Unlike fax, phone, or moving paper documents, electronic documents and data cannot be accessed in transit when encrypted with standard RSA 128-bit encryption methods.7 The individual investigator has no study data or software, so physical security of the site computer is not necessary.

In response to growing national concern about the quality of medical care and patient safety issues, the need for outcome research is obvious. The web provides a natural environment for the conduct of clinical research, leading to enhanced efficiency, accuracy, subject safety, and investigator confidentiality.

References

1. Portney LG, Watkins MP. Foundations of Clinical Research, Applications to Practice. Norwalk, Connecticut: Appleton & Lange; 1993.
2. Department of Practice Parameters. Outcomes Research Resource Guide. USA: American Medical Association; 1996.
3. Hulley SB, Cummings SR. Designing Clinical Research. Baltimore, MD: Williams & Wilkins; 1988.
4. Kane RL. Understanding Health Care Outcomes Research. Gaithersburg, MD: Aspen Publication; 1997.
5. Committee on Quality of Health Care in America, Institute of Medicine. To Err is Human. Washington, D.C.: National Academy Press; 2000.
6. Marks RG, Conlon M, Ruberg SJ. Paradigm shifts in clinical trials enabled by information technology. Statistics in Medicine (in press), 2000.
7. RSA Inc. Security Protocols Overview. ftp://ftp.rsa.com/pub/pdfs/protocols.pdf.