Cross-Cultural Adaptation and Translation of the Neck Dissection Impairment Index (NDII) into Spanish

Abstract

Importance

Selective, modified radical, and radical neck dissections are common surgical procedures that can result in significant musculoskeletal issues of the neck and shoulder. Quality-of-life evaluations after neck dissection must assess and quantify these dysfunctions to allow subsequent comparison of outcomes after different treatments.

Objective

There is no validated Spanish-language questionnaire designed to evaluate neck and shoulder dysfunction after cervical lymphadenectomy. We therefore sought to translate a version of the Neck Dissection Impairment Index (NDII) into Spanish.

Design, Settings, and Participants

A three-phased approach was used. Phase 1: The NDII was translated from English to Spanish using a “forward and backward” translational technique following international guidelines. Phase 2: The questionnaire was administered to six patients from our head and neck surgery clinic who were bilingual and fluent in both English and Spanish. Phase 3: The final version was administered prospectively to 34 patients with prior history of neck dissection (ND). These patients were asked to complete the questionnaire a second time 3 to 6 weeks after their first response. Test–retest reliability was calculated with Spearman’s correlation. Internal consistency was elicited using Cronbach’s alpha.

Main Outcome(s) and Measure(s)

NDII score at initial administration and follow-up administration 6 weeks later, demographic data.

Results

NDII was successfully translated and validated into Spanish. Cronbach’s alpha revealed high internal consistency at both the first time point 0.95 (mean standardized score: 95 (88.1, 97.5), 95% CI: 0.89, 0.97) and at the second time point 0.90 (mean standardized score: 92.5 (80.6, 100.0), 95% CI: 0.81, 0.95). The Spearman’s correlation for test–retest reliability of overall score was strong (rho = 0.772, P < 0.001). The intraclass correlation coefficient of the overall score was moderate (ICC = 0.683, P < 0.001).

Conclusions and Relevance

NDII is a recognized, previously validated quality-of-life (QOL) tool for the identification of ND-related dysfunction. This validated Spanish version will allow clinicians to adequately assess the neck and shoulder-related QOL for the Spanish-speaking population who are under-represented in head and neck research.

Graphical Abstract.

Introduction

Spanish is the second most common language spoken in the United States with approximately 42 million people speaking Spanish as their first language. ¹ 1.78 million Canadians can speak Spanish, making it the most common foreign language spoken, with a 63.0% growth in native speakers since 2001. ² Within the United States, although people of Hispanic origin make up 19% of the total population, it has been shown that only 8% participate in clinical research. Therefore, the NIH has indicated a need for increased recruitment of minority populations in clinical research. ³

For research conducted in the United States, validating culturally appropriate questionnaires in Spanish is one way of addressing this disparity. This is particularly important in the conduct of multi-institutional cooperative group trials. In fact, our participation in one such trial, the NRG Oncology HN006 (a randomized phase II/III trial of sentinel lymph node biopsy versus elective neck dissection for early stage oral cavity cancer), revealed that there is no Spanish-language version of the Neck Dissection Impairment Index (NDII).

The NDII was developed by Taylor et al. ⁴ and is a validated instrument that assesses shoulder and neck musculoskeletal function as well as activities of daily life related to function.^{5 –7} It is designed particularly to assess the physical and psychological effects of oncological neck surgery and is the only validated instrument specifically designed for this purpose. It has been used to assess shoulder and neck musculoskeletal function in head and neck cancer patients in numerous other studies. ⁸ Peer-reviewed publications exist validating translations of the NDII into Danish, ⁹ Italian, ¹⁰ and Canadian French. ¹¹

The lack of a Spanish-language equivalent questionnaire has prevented the accrual of Spanish-only speakers to the NRG Oncology HN006 funded trial, which has been ongoing for over a year and has accrued several hundred patients. We sought to use similar techniques and approaches as those used in these previous publications to develop a reliable version of the NDII questionnaire for Spanish-speaking patients.

Materials and Methods

This study was performed in accordance with the ethical standards of the National Research Committee and with the 1964 Helsinki Declaration and its later amendments. It was submitted and approved by the Internal Review Board (IRB) at our tertiary academic center. Permission was granted by contacting the developers at DChepeha@med.umich.edu. Future use of this Spanish version must be granted by contacting FCivanto@med.miami.edu who will obtain joint consent from the senior author.

The Neck Dissection Impairment Index (NDII) is a 10-item questionnaire that uses a 5-point Likert scale to evaluate impacts on QOL after ND. Scoring was achieved by patients rating response items 1 to 5. On the instrument, the highest score (5) represented better QOL related to neck dissection. Raw score was calculated by summating the 10 questions, each with possible values from 1 to 5. The “raw score” ranges from 10 (worst) to 50 (best). The score was transformed to a 0 to 100-point scale to yield 100 for better QOL related to ND and 0 for the worst QOL related to ND, $\left(\frac{\mathrm{RawScore}-10}{40}\times 100\right).$

Spanish Questionnaire Development: Translation and Cross-Cultural Adaptation

Following the recommendations of the International Society for International Society for Pharmacoeconomics and Outcomes Research, the translation of the questionnaire with cultural adaptation was achieved. ¹² The original NDII questionnaire in English (Figure 1A) was translated into Spanish (Figure 1B) by two language professionals of different national origins (Spain/Cuba and Mexico). Two versions were created, and versions were reconciled between translators to achieve a universal version of the translation. Patients completing the questionnaire were instructed to fill in by hand one of five choices: “Nada (5/5),” “Un poco (4/5),” “Cantidad moderada (3/5),” “Mucho (2/5),” and “Muchisimo (1/5).” Higher scores indicate improved quality of life.

Figure 1.

(A) Neck Dissection Impairment Index Questions in English*: Ten-item validated Neck Dissection Impairment Index questions. Respondents answered: “not at all (5),” “a little bit (4),” “a moderate amount (3),” “quite a bit (2),” or “a lot (1).” Standardization for score of 100: $\left(\frac{RawScore-10}{40}\times 100\right).$ (B) Neck Dissection Impairment Index Questions in Spanish**: Ten-item validated Neck Dissection Impairment Index questions. Respondents answered: “not at all (5),” “a little bit (4),” “a moderate amount (3),” “quite a bit (2),” or “a lot (1).” Standardization for score of 100: $\left(\frac{RawScore-10}{40}\times 100\right).$

*There are questions, but these questions are not the official version of the questionnaire (instrument) that is formatted for presentation to patients. To obtain the version for patients, please see the instructions in the methods and materials.

**Están las preguntas, pero estas preguntas no son la versión oficial del cuestionario (instrumento) que está formateado para presentación a los pacientes. Para obtener la versión para pacientes, consulte las instrucciones en los métodos y materiales.

A language professor familiar with the process of instrument validation examined semantic, idiomatic, and conceptual issues to further refine the survey. A final version was obtained and given to translators to produce a literal translation into English. The two translators and an expert committee synthesized the results of the translations in an English back-translated version that was compared with the original for semantic value.

An example of how this process occurred can be provided by discussing the example of question number 9. There was a debate between translators on this question as they tried to find the best translation for “leisure and recreational activity” in question 9. The most exact translation of leisure is “ocio,” but this is used more commonly in Spain and some Latin-American countries, but some do not use the variation at all. We ultimately left this word in, but also used, the word “diversión,” which roughly means “fun activities.” While implying more intense activity than “leisure time,” it is more widely recognized and understood. Other options proposed included “entretenimiento” (entertainment) or “su tiempo libre” (free time). Ultimately, we went with the most exact translation and made a point of explaining this vocabulary to patients, to ensure they were familiar with the word “ocio” during the initial counseling prior to completion of the questionnaire.

Six bilingual patients, with complete fluency in both languages, looked at both versions and took the questionnaire in Spanish and in English. These individuals completed the Spanish questionnaire first followed by the English version. When answers were compared for all participants, the answers and comprehension of their meaning were deemed consistent from Spanish to English when reviewed by a bilingual physician and a bilingual nurse.

One anticipated issue was that bilingual persons tend to have a higher average level of education and a larger vocabulary than those who are monolingual. Therefore, we had three early trial patients who were only Spanish speakers take the questionnaire. They were not included in the final study results but were encouraged to suggest changes prior to the distribution of the questionnaire to the sample cohort. One such change occurred in question 3, where we originally had used the word “aseo” for hygiene but switched to “higiene personal,” which was more recognizable for our group of patients of various Latin-American national origins. Initial modifications all occurred during the lead-in phase as a fluid process, and not every modification was recorded with date and time. However, all 34 patients presented in our materials and methods section did take the final version of the Spanish NDII questionnaire.

During protocol administration, we did note some patients, presumably those with less education, had more trouble understanding words—and required greater grammatical explanation—than our bilingual initial reviewers and lead-in reviewers. Counseling was limited to clarification regarding the definition of specific words, ie “ocio,” and care was taken to ensure the patient independently read and answered the written questionnaire.

Eligibility

Eligibility criteria for enrollment included patients over 18 years of age who had unilateral or bilateral neck dissection involving dissection of the spinal accessory nerve for head and neck malignancy at our tertiary academic institution from 2020 to 2023. Patients were required to have a command of the Spanish language with a strong preference to utilize Spanish in daily conversation. They had to be able to read, comprehend, and write in Spanish. They also needed to be able to understand and agree to the consent criteria and sign the consent form. This was confirmed by patient self-identification and through interaction in Spanish with the Spanish-speaking physicians and nurses in the study.

Patients were excluded if a review of their medical record and patient report indicated that they had undergone surgery less than 3 months ago or more than 3 years ago, and had cancer recurrence, unrelated neck or shoulder pathological disease, or neuromuscular disease. ¹³ They also were excluded if they had a lack of basic written and oral command of the Spanish language, as self-reported by the patient and confirmed by one of three Spanish-speaking investigators on the initial interview at the time of potential consent.

As per IRB requirements, patients were offered participation in the translation trial only at the time of previously scheduled appointments in the head and neck clinic, and the first questionnaire was completed in person at the time of the visit, after appropriate counseling by a physician or nurse investigator. The second questionnaire was completed in person or via live video conference. Based on a desire to come as close as possible to the “in-person” interaction, phone conversation without video was not permitted. For the second questionnaire, the interviewer was also a physician or nurse investigator, and counseling was again restricted to vocabulary, especially reviewing certain key vocabulary points, as mentioned above.

Neck dissections were required to include dissection of level 2-4, with or without level 5, according to the neck dissection classification of the American Head and Neck Society and the American Academy of Otolaryngology-Head and Neck Surgery ¹⁴ so all participants had manipulation of the spinal accessory nerve.

Validation of Questionnaire

To evaluate the test–retest reliability of the NDII, the questionnaire was administered twice, 3 to 6 weeks apart, to all patients enrolled. This interval period was selected because no substantial change was expected to take place in the subjects’ condition within this short period, but the period was long enough so that the patient would probably not remember specifically their answers from the first time they completed the questionnaire.

When completing the second NDII, the subjects did not have the chance to check their responses on the previous questionnaires, nor were they given any reminder regarding the previous completion of the questionnaire. They were counseled at the time of completion and assisted by either a physician on the study or a nurse on the study, only for clarification of questions regarding the definition of words. The health care provider counseling and assisting them followed similar rules for counseling on the first and second events. For patients completing the questionnaire the second time, the health care provider did not review the prior questionnaire or prompt the person completing the questionnaire regarding their prior answers.

Statistical Analysis Methods

The Likert scale of each item is treated as ordinal data, where an overall measurement of the NDII questionnaire is summed up by all items as an overall score. To examine the internal consistency of the translated NDII questionnaire, Cronbach’s alpha was implemented. The Shapiro–Wilk test is used to check the normality of data distribution. The Spearman’s rank correlation coefficient was used to assess the test–retest reliability of each item and overall score between both administrations due to the ordinal nature of the data and its non-normal distribution. To further assess the robustness of the results, Pearson’s product-moment correlation coefficient was also calculated for comparison, acknowledging its assumption of linear relationships and interval data. In addition, the Intraclass correlation coefficient (ICC) was calculated to evaluate the degree of absolute agreement between test and retest scores, providing a more comprehensive measure of reliability that accounts for both systematic and random variations. The consistent findings across these approaches strengthen confidence in the reliability estimates. According to guidelines by Akoglu ¹⁵ and Koo and Li ¹⁶ , a Pearson’s or Spearman’s correlation coefficient greater than 0.7 or less than −0.7 is considered a strong correlation. An ICC value between 0.5 and 0.75 indicates moderate reliability. Any P-value under 0.05 is considered statistically significant.

All the data analysis is performed by R studio (R version 4.4.1). ¹⁷

Results

Demographic data was collected with clinical and pathological tumor stages. Additional data considered included age, sex, national origin, and surgical procedure (selective neck dissection vs modified radical or radical neck dissection) is included in Tables 1 and 2.

Table 1.

Demographic Data.

National origin (n = 34)	Frequency (%)
Cuba	21 (61)
Venezuela	3 (9)
Colombia	2 (6)
Dominican Republic	2 (6)
Argentina	1 (3)
Guatemala	1 (3)
Mexico	1 (3)
Peru	1 (3)
Puerto Rico	1 (3)
Missing	1 (3)
Age (n = 34)
30-39	1 (3)
40-49	4 (12)
50-59	12 (35)
60-69	9 (26)
70-79	5 (15)
80-89	3 (9)
Sex (n = 34)
Female	10 (29)
Male	24 (71)
BMI (n = 34)
16-20.99	3 (9)
21-25.99	15 (44)
26-30.99	11 (32)
31-35.99	3 (9)
36-40.99	1 (3)
41-45.99	0 (0)
46-50.99	1 (3)

Table 2.

Interventions Performed.

Radiation (n = 34)	Frequency (%)
None	12 (35)
Preoperative	3 (9)
Postoperative	15 (44)
Preoperative and postoperative	4 (12)
Operation (n = 34)
Selective Neck Dissection, 1-4	14 (41)
Selective Neck Dissection, 2-4	18 (53)
Modified Radical Neck Dissection	1 (3)
Radical Neck Dissection	1 (3)

Men were more common than women to undergo a neck dissection, likely due to the greater frequency of squamous cell carcinoma in men. Since we included multiple different histological diagnoses, we did not evaluate this difference statistically. No detectable difference in demographics was found between the variety of neck dissection performed.

18 patients received selective neck dissection (SND) of levels II through IV, 14 received SND of levels I through IV, and 1 received a modified radical neck dissection with dissection of levels I through V. The spinal accessory nerve was dissected but preserved in all these cases. One patient received a radical neck dissection with a sacrifice of the spinal accessory nerve. ¹⁰ Nine patients had bilateral neck dissections, all selective.

34 patients completed both copies of the translated questionnaires. Internal consistency proved to be excellent at both administrations, with Cronbach’s alpha of 0.95 [95%CI 0.89, 0.97; mean standardized score: 95 (88.1, 97.5)] at the first time point and 0.90 [95%CI 0.81, 0.95; mean score: 92.5 (80.6, 100.0)] at the second time point (Figure 2, Table 3). This suggests that the items in the translated NDII questionnaire reliably measure the same construct. The test–retest reliability by Spearman’s rank correlation is strong (rho = 0.772, P < 0.001), demonstrating that the scores from the first and second administrations are highly consistent and stable over time. The intraclass correlation coefficient was moderate for overall test agreement (ICC = 0.683, P < 0.001, Table 4).

Figure 2.

Linear Frequency Plot of First and Second Administration: Linear frequency plot representing the first test administration (blue) and second test administration (orange) given six weeks apart. No significant difference was found between the two administrations (P = 0.747).

Table 3.

Neck Dissection Impairment Index Scores (Test and Retest Values).

	Administration #1	Administration #2	P-value
Question 1	4 (4-5)	4 (4-5)	0.82
Question 2	4.5 (4-5)	4 (4-5)	0.85
Question 3	5 (5-5)	5 (5-5)	0.99
Question 4	5 (5-5)	5 (5-5)	0.43
Question 5	5 (4-5)	4.5 (4-5)	0.99
Question 6	5 (4-5)	5 (4-5)	0.33
Question 7	5 (4-5)	5 (4-5)	0.34
Question 8	5 (5-5)	5 (5-5)	0.85
Question 9	5 (5-5)	5 (5-5)	0.89
Question 10	5 (4.25-5)	5 (4-5)	0.73
Raw Total	48 (45.25, 49.0)	47 (42.3, 50.0)	0.61
Standardized Score*	95 (88.1, 97.5)	92.5 (80.6, 100.0)	0.61

^*Standardization for score of 100: $\left(\frac{RawScore-10}{40}\times 100\right).$

Table 4.

Test–Retest Reliability for Questionnaire Administered at Two Time Points.

Test–retest reliability
Items	Pearson’s product-moment correlation coefficient		Spearman’s rank correlation coefficient		Intraclass correlation coefficient
	r (95%CI) a	P-value d	rho (95% CI) b	P-value d	ICC (95% CI) c	P-value d
Q1	0.629 (0.379, 0.794)	<0.001	0.646 (0.404, 0.804)	<0.001	0.608 (0.354, 0.778)	<0.001
Q2	0.711 (0.499, 0.843)	<0.001	0.751 (0.562, 0.866)	<0.001	0.715 (0.507, 0.844)	<0.001
Q3	0.417 (0.103, 0.656)	0.011	0.269 (−0.066, 0.549)	0.113	0.409 (0.097, 0.648)	0.006
Q4	0.319 (−0.010, 0.586)	0.058	0.464 (0.160, 0.688)	0.0043	0.308 (−0.018, 0.575)	0.0315
Q5	0.608 (0.349, 0.781)	<0.001	0.574 (0.302, 0.759)	<0.001	0.609 (0.351, 0.780)	<0.001
Q6	0.653 (0.413, 0.808)	<0.001	0.617 (0.361, 0.786)	<0.001	0.644 (0.405, 0.801)	<0.001
Q7	0.670 (0.438, 0.818)	<0.001	0.660 (0.424, 0.813)	<0.001	0.617 (0.367, 0.784)	<0.001
Q8	0.667 (0.433, 0.817)	<0.001	0.806 (0.649, 0.897)	<0.001	0.654 (0.417, 0.808)	<0.001
Q9	0.343 (0.016, 0.604)	0.041	0.588 (0.322, 0.768)	<0.001	0.335 (0.006, 0.596)	0.0232
Q10	0.282 (−0.051, 0.559)	0.095	0.584 (0.316, 0.766)	<0.001	0.288 (−0.048, 0.563)	0.0453
Total	0.693 (0.472, 0.832)	<0.001	0.772 (0.595, 0.878)	<0.001	0.683 (0.459, 0.825)	<0.001

^aCorrelation coefficient of Pearson’s product-moment correlation.

^bCorrelation coefficient of Spearman’s rank correlation.

^cIntraclass correlation coefficient.

^dAny P-value under .05 is considered statistically significant.

^*Bolded values are statistically significant

Discussion

A national cooperative group study, NRG Oncology HN006: Randomized phase II/III trial of sentinel lymph node biopsy versus elective neck dissection for early stage oral cavity cancer, has been unable to invite the participation of Spanish-speaking patients, due to the lack of a validated NDII equivalent in Spanish. The Spanish translation tested here can be used as a tool to expand eligibility to Spanish-speaking patients for clinical research in the field of head and neck surgery, including in the ongoing HN006 trial.

Some challenges and limitations we encountered as we sought to complete this trial. Due to the length of time between the administration of both questionnaires, 20 patients were lost to follow-up. Though socioeconomic data was not collected, our impression was that the group of patients who speak only Spanish while living in the United States tend to include more patients from lower socioeconomic living situations, while Hispanic patients who speak English equal to or better than Spanish presumably may have greater economic means and higher social status.

We encountered many patients who had limited access to the internet and computers, and difficulty faxing or emailing questionnaires. On three occasions, we had to have patients come in to see us in person or had to meet them at other appointments at the hospital, to achieve completion of the second questionnaire within the specified time frame. Due to the non-normal distribution and small sample size, we did not have sufficient power to detect significant correlations for every item on the survey. This could be the reason for disagreement where two methods arise. Based on the current sample, the total score does suggest a strong correlation for both methods which could be strengthened through a greater sample size.

We did encounter multiple episodes of difficulty related to question 9, and the previously mentioned issue with the word “ocio.” We noticed many patients had trouble understanding the word “ocio,” despite its pairing with a relatively synonymous term, and early on we identified a need to briefly include an explanation regarding the word. We used the term “tiempo de diversion” or “actividades hechas en tiempo libre” (“fun time” or “activities done in your free time”) to ensure the message was understood and made this a standard part of our brief counseling. We also found that it was important to remind patients that we were interested in symptoms only during the last 4 weeks. Otherwise, some patients would go back to reporting the symptoms they had in the two weeks after the initial surgery. With initial attention to these brief counseling issues, we found patients could complete the 10-item questionnaire quickly and easily.

Our results confirm reliability, consistency, and stability between our two points in time, as presented in Table 4, and question 9 remained statistically robust. Our statistically weakest question was question 3, which was not identified as a problem during the lead portion of questionnaire development, possibly because the bilingual group used for lead-in, and even our three initial volunteers, may not be representative of the entire group in terms of educational level. This has not been directly assessed but represents one possible explanation. We suspect that the P-value for this question would improve with larger numbers of patients, but modifications of this translation could be considered in the future. The complete questionnaire was statistically consistent as described above.

Conclusion

The NDII was successfully translated into Spanish and its use was easy for patients. The Spanish translation of the NDII can represent a useful tool for individual patient assessment and future research.