Validation of the Caregiving Knowledge Questionnaire-My: Measuring Knowledge Regarding Positioning and Feeding among Malaysian Informal Caregivers of Stroke Patients

Ethical consideration and informed consent

All participants received a briefing as to the purpose of the study and written informed consent was obtained. All identifying data were replaced with participant numbers during analysis. The study received ethical approval from the Institutional Medical Research and Ethics Committee (FF-2017–196).

Translation and adaptation of the questionnaire

The original survey items were reviewed by the research team for grammatical errors. The initial content review also led to the two items for range of motion exercises to be removed from the questionnaire. This was because upon discussion with a trained physiotherapist and search for guidelines, the questions were deemed invalid since there is no evidence-based recommendation about the ideal frequency of range of motion exercises for stroke patients.[6] The locations of range of motion exercises would also depend on the location of the disability, and hence it would be inappropriate to assess the knowledge of caregivers on locations of range of motion exercises if the patient they cared for did not have disability at certain joints.

Hand drawings replaced the original photographs to ensure that the pictures could be easily reproduced in black and white and also to relabel the affected limbs. This also minimizes ethical issues related to photographs and would be more acceptable in a Muslim-predominant country. Hand drawings are also easier and cheaper to reproduce compared to photographs. The corrected version was then sent to be translated by two independent translators to obtain the Malay version. The Malay version was reviewed by the research team members and reconciled to produce a harmonized version. The research team members were all competent users of Malay and English language; five of the team members were native speakers of Malay. Differences in the terms were reconciled through discussions and final consensus. This harmonized version was subsequently sent to another two independent translators who translated it back into English. Finally, all language versions were compared by the research team members with the original version to ensure semantic and conceptual equivalence.

The harmonized Malay version was then pretested among five caregivers of stroke patients. Pretesting usually requires between 5 and 15 samples and are best done in a heterogeneous population.[7] Further modifications were made, based on pretesting results, to improve the flow of the language and explanatory phrases were added where necessary. Some local cultural examples were given. For example, an answer option to the item on the correct consistency of milk “like yogurt drink” was elaborated by giving examples of popular local brands of yogurt drinks. The amended version was pretested on another two participants and was approved for the validation phase.

There were a total of thirty items in the patient positioning subscale, 25 items in the feeding subscale, and 9 items in the pressure sore prevention subscale. Separate sections could have more than one correct response. Hence, each correct answer was awarded 1 point. Respondents were also awarded marks for not selecting incorrect items. No marks were deducted for wrong answers. Respondents who were not involved in feeding the patient through a tube were exempted from answering the feeding subscale but were required to complete all other sections. Table 1 shows the breakdown of items in the questionnaire used for validation.

Table 1 Original questionnaire items as adapted from Lee et al. (2015)

Validation procedure

In view that there were no prior gold standard questionnaires to allow concurrent validation, we used contrasting groups validation method. In contrasting groups' validation, the samples for each group are purposively recruited to form a priori groups.[8]

The developed questionnaire was tested on 13 caregivers of stroke patients (ages 30–74 years, 9 females and 4 males) using convenience sampling from a long-term stroke clinic. The questionnaire was also self-administered to 10 4^th year medical students (8 females and 2 males, aged between 24 and 26 years) and ten primary care doctors (8 females and 2 males, aged between 30 and 38 years, and who were in their 2^nd and 3^rd years of training in family medicine).

Internal consistency and contrasting groups' discrimination analyses

The internal consistency was determined for each predetermined subscale, namely, patient positioning, feeding, and pressure sore prevention. Items with no variance or had poor item-total correlations were removed. Items that improved the internal consistency when deleted were also removed one by one until a satisfactory internal consistency was achieved. The subscale for pressure sore prevention, however, failed to achieve a satisfactory internal consistency due to the presence of too few items, and thus the subscale was removed from the questionnaire.[9]

The total scores for the subscales were calculated based on the final 29 items for the patient positioning subscale and 15 items for the feeding subscale. The contrasting group method was used to determine the questionnaire's construct validity. Theoretically, informal caregivers should have lower knowledge compared to medical students and doctors. Hence, a significant difference in scores between these three groups would support the tool's ability to discriminate between good and poor knowledge in this domain.

Patient positioning subscale

The initial Cronbach's alpha for this subscale was 0.637 [Table 2]. From the initial 30 items, items Q2a and Q4b were dropped. Item Q2a referred to the placement of a pillow under the patient's knees while lying supine, whereas Q4b referred to use of an arm sling while lying supine. Theoretically, removing items from each question may affect the difficulty level of the questions as the number of possible options is reduced. It is also important to retain correct positioning options. Dropping items Q2a and Q2b did not remove the correct answers and kept at least two items to be considered by the participants.

Table 2 Internal consistency of the subscales

Feeding subscale

The initial Cronbach's alpha was 0.544 for the preliminary 24 items. Items with poor item-total correlation were removed one by one to improve the internal consistency. The final Cronbach's alpha for the feeding subscale was 0.695 after dropping five items. Again, before removing the items, care was taken to ensure that items representing correct feeding knowledge were not inadvertently removed from the questions. Each question also had at least three options to choose from to preserve its difficulty index.

Pressure ulcer prevention subscale

Cronbach's alpha for the pressure ulcer prevention subscale was extremely low (0.327). This may have been due to an insufficient number of items in the subscale.[9] To improve the subscale, it is thought that more items related to pressure ulcer prevention would need to be developed and tested. Hence, the subscale for pressure ulcer prevention was removed from the questionnaire in the current study.

Contrasting groups validity

The scores for each subscale were summed. The overall mean score for the patient positioning subscale was 20.30 ± 3.72, whereas overall mean score for the feeding subscale was 10.57 ± 2.30. The scores of both subscales were normally distributed and are displayed in Table 3. Overall, there was a significant difference in mean scores between the different groups.

Table 3 Mean scores by respondent group for each subscale

Post hoc tests [Table 4] revealed significant mean differences between caregivers' scores and doctors' scores for both the patient positioning and feeding subscales. Caregivers' scores for feeding were also significantly lower than medical students' scores. Medical students' and doctors' mean scores for feeding were comparable.

Table 4 Post hoc analyses for subscales by respondent groups

Financial support and sponsorship

Partial funding was received from the Universiti Kebangsaan Malaysia (Grant No: GGP-2017-019). No further funding from the public, commercial, or not for profit sectors was received.

Conflicts of interest

There are no conflicts of interest.