LITERATURE REVIEW

TOP ABSTRACT LITERATURE REVIEW MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Computer-administered HHs have been widely studied in a variety of medical settings, and the specific advantages of using them have been identified. Use of a computer has been shown to be an efficient and a reliable way to collect health information, with 85 to 97 percent reproducibility.^1–3 In a study of 2,424 obstetric patients, researchers found that the computer was better in eliciting information about patient symptoms than was a self-reported paper questionnaire.⁴ They found, however, that structured paper questionnaires did better than the computer in gathering information about obstetric, medical and surgical histories and treatments.

The computer also has been found to be more successful in eliciting information on sensitive issues than have face-to-face interviews or paper questionnaires. Researchers found that computer-generated questions regarding lifestyle,⁵ risk factors associated with gynecologic problems⁶ and HIV status^7,8 were answered more forthrightly than were questions in paper questionnaires or interviews.

Researchers have consistently noted that patients found computer questionnaires easier to use than paper ones and preferred using the computer over paper.^8–10 A blood donor study found that subjects "enjoyed" the computer interview and judged it to be more private than the standard method of donor assessment.⁷ At the National Institutes of Health in Bethesda, Md., a portable, patient-interactive computerized system was developed to obtain medication histories modeling pharmacist-conducted medical-history interviews.¹⁰ The questions were written in lay language, and the patients easily and accurately entered data.

Researchers in the Nether-lands studied the applicability of a computerized medical history at an endoscopy unit of a university hospital.² Average completion time for both paper and computer questionnaires was about 11 minutes. After completing the computerized questionnaire, 45 percent of the patients could put their complaints into words more easily. These researchers concluded that the computerized questionnaires provided a detailed and uniformly acquired medical history of most patients.

The clinical use of computers to collect and manage dental patient information is minimal.

Dental patients do not always provide accurate or complete health information. One study found that one-third of questionnaires contained incorrect or missing data.¹¹ A study of 107 patients at a university dental hygiene, or DH, program clinic found that a significant number of patients provide inaccurate or incomplete information to questions routinely asked on the HH form.¹² The reasons for inaccurate or incomplete HHs are many. Patients frequently do not have the opportunity to or are too timid to ask for explanations from the dental team. The patient may not understand the terminology,² or may discount the importance of giving medical data for dental treatment,¹³ fail to recall important facts at the time the questionnaire is completed,¹⁴ judge certain conditions as "unimportant"¹⁵ or hesitate to answer sensitive questions.

Dental health care providers are steadily tapping into the power of the computer to improve the quality of care they provide. The clinical use of computers to collect and manage dental patient information, however, is still minimal. We found no published research on the use of a computer to collect HHs directly from dental patients. Patients’ completion of a computerized medical or dental history questionnaire offers many potential benefits:

– patients cannot leave questions unanswered;

– patients may respond more forthrightly;

– explanations and examples can be provided;

– definitions can be provided;

– patient understanding may be increased;

– less time is required to complete the questionnaire;

– patient responses are legible;

– data entry step for computerized patient record is eliminated.

	MATERIALS AND METHODS

TOP ABSTRACT LITERATURE REVIEW MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
We used the Automated Personal Health Inventory for Dentistry, or APHID, which was programmed in the C++ for Windows environment to use with a hand-held, pen-based computer (Epson EH-400, Seiko Epson Corp.; Windows for Pen, Microsoft Corp.) with a touch-sensitive display and a stylus for user input. We chose the pen-based computer so subjects could use a stylus rather than a keyboard or mouse. We believed that using a stylus is more natural and familiar to novice computer users.

The HH questionnaire had 78 questions, which appeared sequentially and one at a time on the computer screen (Figure 1). The subjects answered the questions by tapping the stylus on the screen over the button representing the response to the question. The choice of answers was "Yes," "No" and "I don’t know." A large icon—a question mark superimposed on a drawing of a tooth—was located in the upper left-hand corner of the screen. Subjects who did not understand the question or who wanted more information about it could tap on the question mark icon to see an explanation or definition, as well as examples of what was meant by the question (Figure 2). (Explanations for each item on the HH were written by the dental hygienist author.) The bottom one-half of the display contained a graphic appropriate to the item (Figure 1), as well as a progress bar that estimated how close subjects were to being done. Subjects also could go back to a previous question if they wished to do so. APHID automatically captured the length of time a patient took to complete the HH questionnaire, how many times a patient asked for definitions or explanations by tapping on the question mark icon, and which specific HH items he or she queried.

View larger version (25K): [in this window] [in a new window]   Figure 1. View of Automated Personal Health Inventory for Dentistry with a graphic that correlates with the question being asked.

View larger version (38K): [in this window] [in a new window]   Figure 2. View of Automated Personal Health Inventory for Dentistry showing an explanation of a question.

We obtained subjects from two different settings: new patients at a DH clinic that is part of a baccalaureate DH program at UMC, and people who were friends or relatives waiting with patients at an outpatient medical clinic at UMC. After reading a brief letter explaining the project, we orally reviewed the subjects’ proposed involvement, and subjects were given an opportunity to ask questions and practice using the stylus with the computer.

All subjects at the DH clinic agreed to participate in the study. Only two people at the medical clinic declined to participate. Subjects at the DH clinic signed a consent form to participate in the study, as their paper HH questionnaires were part of their official dental charts and the data they provided were identifiable. Subjects at the medical clinic did not sign consent forms, as the data they provided were anonymous. All subjects completed the HH questionnaire in both formats—paper and APHID. The control group completed the paper form first, and the experimental group completed the computerized version first. We alternately assigned the subjects to the control or experimental groups.

After completing the two HH questionnaires, subjects completed an anonymous short paper survey designed to elicit their opinions on and attitudes about the two HH collection methods. This survey also asked for brief demographic information about the subjects, as well as their experiences with technology. We estimated the subjects’ level of comfort with using computer technology by whether they used an automated teller machine and/or a computer at home or work.

	RESULTS

TOP ABSTRACT LITERATURE REVIEW MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Fifty-two subjects participated in the study. We were not able to include two subjects in our analysis, since the last page on each of their paper questionnaires was not completed; the subjects probably did not realize that there was a fourth page. Therefore, 50 subjects completed all parts of the study. Fifty-one subjects completed opinion surveys, but because the opinion surveys were completed anonymously, we could not identify which one corresponded to which HH. Thus, we had 51 opinion surveys and 50 sets of HH questionnaires. Twenty-two subjects were from the DH clinic (nine in the control group and 13 in the experimental group), and 28 subjects were from the medical clinic (10 in the control group and 18 in the experimental group). Therefore, 19 subjects were in the control group, and 31 were in the experimental group. The subjects’ demographics are summarized in Table 1, and Table 2 provides data on the subjects’ experience with technology by age, sex and education.

View larger version (50K): [in this window] [in a new window]   Figure 3. Subjects’ opinions of the Automated Personal Health Inventory for Dentistry.

The reliability of the questions by category were: general health (90 percent), medications (96 percent), review of body systems (94 percent), sensitive questions (95 percent) and dental questions (89 percent).

An unanticipated result we found was that only two of the 50 subjects (4 percent) answered all 78 questions on the HH questionnaires exactly the same using both paper and APHID methods. The mean number of inconsistent answers per subject was 5.6. Figure 4 shows the frequency of subjects by number of inconsistent answers. We did not attempt to determine which answers were actually correct. The 28 subjects from the medical clinic waiting room completed the HH questionnaires anonymously, and we did not conduct face-to-face interviews with them to discuss their answers. The 22 subjects from the DH clinic completed the HH questionnaires as part of a scheduled appointment, and the DH student providing care interviewed the subjects. The subjects’ two HHs had not been compared before this interview, so any inconsistencies were not evident to the DH student. Because of the small number of subjects from whom a correct answer could be determined, we did not attempt a statistical study of the inconsistent answers.

View larger version (49K): [in this window] [in a new window]   Figure 4. Frequency of conflicting answers between the health histories.

Sixty-four percent of the subjects answered all of these nine questions consistently. Data from the 36 percent of subjects (18 subjects) who answered inconsistently indicate that some subjects answered sensitive questions more forthrightly using the computer. For example, two subjects admitted to having a history of mental problems on the computer questionnaire, but answered "No" on the paper questionnaire. Four subjects admitted they did not floss every day on the computer questionnaire, but indicated on the paper questionnaire that they did. Four subjects said they did not know if they used recreational drugs on the computer questionnaire but on the paper questionnaire answered that they did not. On the computer questionnaire, one subject admitted that she had fainted while undergoing dental treatment but did not admit to it on the paper questionnaire. None of the subjects indicated having a history of sexually transmitted disease or acquired immunodeficiency syndrome on either the paper or computer questionnaires.

The APHID asked the subjects, "Do you use alcohol?" If they answered "yes," they next were asked to check one of four levels of use: special occasions, a few times a month, a few times a week or about every day. On the paper questionnaire, however, we presented them with the four levels of use without first asking, "Do you use alcohol?" Six female subjects responded differently to the question of alcohol use. These subjects checked on the paper questionnaire that they used alcohol on special occasions but answered "no" when asked on the APHID, "Do you use alcohol?" Apparently, these did not believe that the use of alcohol on special occasions constituted use of alcohol.

Use of explanation function. Only one subject tapped the question mark icon to receive an explanation of a computer questionnaire item. Therefore, we were unable to determine whether providing an explanation of the item would result in a more accurate HH. Given the length of the HH, the subjects may have been more concerned about finishing it and did not take time to ask for explanations.

	DISCUSSION

TOP ABSTRACT LITERATURE REVIEW MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
The subjects in this study reported that they believed the computer questionnaire was easier to use than the paper questionnaire; these results are similar to studies discussed previously in the literature review. APHID presented the questions one at a time and in a larger font, which may have improved the readability and ease of use in the subjects’ minds. Similarly to subjects in the study of blood donors,⁷ our subjects said the computer method was more private and took less time than the paper method (Figure 3).

Fifty-six percent of the subjects left at least one item unanswered on the paper questionnaire. Therefore, the APHID HHs were more complete than the paper HHs since the computer required a response to every question. We would like to note, however, that complete does not necessarily mean accurate.

The font and format of the paper questionnaire could be partially to blame for missing and inconsistent answers. The paper questionnaire consisted of three and one-half single-spaced pages with narrow side margins. Some check-box items were embedded in the text and could be missed easily. The font—a nonproportional, typewriter style—generally is more difficult to read.

Another difference between the two types of questionnaires was that the items in the paper questionnaire were surrounded by other questions and subheads, which could have provided context to the question. In APHID, the sequence of questions was identical to that on the paper questionnaire, but only one question was visible at a time. For example, on the paper questionnaire, the question "Have you ever had or been treated by a physician for swollen ankles?" was found with other questions about heart symptoms such as chest pain and shortness of breath. One subject in the experimental group answered "Yes" to this question on the computer questionnaire. He then answered "no" to swollen ankles on the paper questionnaire, realizing by the context that swollen ankles referred to heart problems rather than the multiple sprained ankles he had experienced.

Many of the items that were answered unreliably in this study probably would not have a serious impact on patient care. An overall 93 percent reliability rate is admirable; however, a 7 percent inconsistency rate could cause harm if not recognized by the clinician. For example, hypersensitivity to latex and/or medication allergy could have serious consequences.

This study revealed a high level of acceptance of the patient-completed pen-based computerized health history questionnaire.

	CONCLUSION

TOP ABSTRACT LITERATURE REVIEW MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSION REFERENCES

 
Further research is needed to determine the cause of inconsistent answers. If subjects completed the APHID questionnaire twice, would there be better reliability than between the APHID and paper methods? In other words, is the questionnaire’s format the cause of unreliability, rather than the collection method used? Would an easier-to-read paper version increase reliability? The problematic questions on the questionnaire should be rephrased to improve clarity.

As only one subject used the explanation feature on the computer questionnaire, we were unable to test our hypothesis that providing explanations can increase accuracy. One option for future study would be to automatically display the explanation for every item; however, there would be no way to determine whether a subject actually read the explanation. Another disadvantage could be that subjects may become impatient or overwhelmed if they conclude they must read everything on the screen. Another possible way to provide explanations is to embed the help icon at the end of the question or after the key term in the question to remind subjects that help is available. Our design required the user to click on the icon in the upper left-hand corner of the computer screen.

This pilot study revealed a high level of acceptance of the patient-completed pen-based computerized HH questionnaire. Patients can easily use a computer that does not require use of a keyboard or mouse, and they reported that it took them less time to complete the computer questionnaire than the paper questionnaire. Patient entry on a pen-based computer also provided an easy way to input the HH data into a computerized patient record system; importing the data rather than having a clerk enter the data from a paper HH is more accurate and efficient.

We believe that a large sample study needs be conducted to research the problem of inconsistent answers. In the study, the HH data should be collected twice, and the answers should be compared for inconsistencies before the subjects are interviewed. In ordinary practice, only a single HH is completed, so a health care professional has no way to determine if an answer is inconsistent unless it becomes apparent in the follow-up interview.

Computer software could analyze the responses and provide a summary to the practitioner that identifies the items that definitely should be included in the HH review. Software also could identify inconsistent answers that need clarification—for example, a patient who indicates he is not diabetic but is taking an antidiabetic drug. This would allow the professional to do an efficient and thorough review of items relevant to each patient. Finally, regardless of the method used to collect the HH, it is imperative that the dental health professional review the information with the patient.