Reliability, validity, and minimal clinically important differences for the Thai-version of the Aberdeen Varicose Vein Questionnaire (AVVQ-Thai) in patients with chronic venous disease

Participants

The study included adults of ≥18 years of age able to read Thai and Clinical-Etiology-Anatomy-Pathophysiology (CEAP) C2 to C6 with symptoms related to the venous disease and unilateral pathologic reflux of great (GSV) or small saphenous veins (SSV) who would undergo venous intervention. Exclusion criteria were concomitant peripheral arterial disease, morbid obesity (body mass index ≥ 35 kg/m²), pregnancy, recurrence varicose vein, history of previous venous intervention, and history of deep vein thrombosis or presence of deep vein obstruction from duplex scan. All participants signed informed consent before entering the study.

Aberdeen Varicose Vein Questionnaire (AVVQ)

AVVQ is a clinically derived questionnaire specifically developed for patients with varicose veins. First developed and tested for measurement properties by Garratt et al. in 1993,³ it comprises four domains (pain and dysfunction, cosmetic appearance, extent of varicosity, and complications) with 13 questions that are summed based on a clinical weighting and scored from 0 to 100, where higher score represent greater CVD severity.^3,11

Translation process (Figure 1)

Figure 1. Forward and backward translation process of the AVVQ.

SW and a team form the Language Institute of Thammasat University undertook translation which followed recommendations by the European Organization for Research and Treatment of Cancer.¹⁶ First, the English version of AVVQ was independently translated into Thai by two professional Thai translators, trying to make the language clear and as simple as possible. Then, the research team assessed both translations for equivalence with the original one. Any discrepancies were resolved by consensus and resulted in a single provisional forward translation. After that, the provisional Thai forward translation was back translated to English by another native English speaker and compared with the original English version to ensure equivalence. The final translated Thai version was reviewed by the research team and was used in the study. The English translated and final translated Thai versions were uploaded online as extended data and were available at https://doi.org/10.5281/zenodo.10990113.¹⁵

Main survey and outcome measurement

Baseline characteristics, EQ-5D-3L, preoperative AVVQ-Thai, and venous clinical severity scores (VCSS), were obtained at an initial outpatient visit using a self-completed pen and paper questionnaire and clinician completion for VCSS. For purposes of assessing test-retest reliability, patients took a second AVVQ-Thai questionnaire and reply-paid envelope home for completion 1 to 2 weeks later (retest AVVQ-Thai). After the venous intervention, postoperative complications, EQ-5D-3L, postoperative AVVQ-Thai, and VCSS were collected one-month post-intervention. Mean scores of the preoperative and retest AVVQ-Thai were used as preoperative scores in the analysis. In the case of missing data for individual AVVQ questions, data from the preoperative or retest questionnaires were substituted accordingly.

Interventions

Refluxed veins were ablated using either endovenous thermal ablation (i.e., radiofrequency (RFA), endovenous laser ablation (EVLA)), or surgical therapy (saphenofemoral ligation with stripping for GSV or SPJ ligation for SSV) depending on patients’ preferences and costs incurred. All procedures were performed under regional anesthesia. Briefly, radiofrequency ablation was performed with tumescent anesthesia to ablate the GSV at 2–3 cm distal to the saphenofemoral junction to the GSV around the knee level. In surgical therapy, saphenofemoral junction was ligated flush to the common femoral vein. Then, GSV was stripped from proximal to distal GSV around the knee level. Pathologic perforator insufficiency was treated by ultrasound-guided foam sclerotherapy in CEAP C5-6. Varicosities would be treated concomitantly by either phlebectomy or foam sclerotherapy.⁴

Statistical analysis

Sample size is estimated using the rule of thumb for questionnaires testing. At least 10 subjects were necessary for 1 question with a minimum number of 100 subjects to ensure the stability of the variance-covariance matrix.¹⁷ As a result, with 13 questions in AVVQ so 130 patients were necessary. Accounting for 10% loss follow up rate, 145 patients would be recruited.

Cronbach’s alpha and intraclass correlation coefficient (ICC) were used to assessed internal consistency and test-retest reliability of the AVVQ-Thai, respectively. Cronbach’s alpha reflects the homogeneity of the AVVQ questions in assessing the same concepts. Higher values reflect higher levels of consistency/homogeneity with level of 0.70 and above being satisfactory.¹⁸ ICC and 95%CI were estimated by a two-way mixed effect model with an absolute agreement for the overall scores and within each item of the AVVQ. Values of less than 0.5, 0.5 – 0.75, 0.75 – 0.9, and greater than 0.9 indicate poor, moderate, good, and excellent reliability, respectively.¹⁹

Hypothesis testing was used to assess the construct validity of the AVVQ-Thai including correlations of the AVVQ scores and VCSS both pre- and postintervention, and comparing median AVVQ scores between patients with different degrees of problems reported each dimension of EQ-5D (no problem vs. some and severe problems). It was hypothesized that AVVQ scores would have low to moderate correlations with the VCSS because while there is some content overlap, the latter is completed by the clinician and is more concerned with symptoms than how varicose veins affect quality of life. It was hypothesized that patients with higher scores on the AVVQ would report more problems across the five EQ-5D dimensions. Responsiveness of the AVVQ-Thai was assessed by determining the change in the AVVQ scores after intervention according to the CEAP clinical class (C2-C3 and C4-C6). Ceiling effects, or the proportion of patients with the best possible scores that cannot be improved with intervention, were also compared for the AVVQ and EQ-5D.

Distribution-based methods were used to estimate minimal clinically important differences (MCID) of the AVVQ.^20,21 There are several methods of MCID derivation. Four types of methods were analyzed and reported for consideration. 1) 0.5 of SD of the observed change (SD_{observed change}) in AVVQ as recommended following a systematic review by Norman et al.²² 2) Standard error of measurement (SEM), which represents the variability of the scores. Any score difference below the SEM is likely due to measurement error rather than genuine change. Thus, it is the most conservative value of MCID. SEM was calculated by the equation $\mathit{SEM}=\mathrm{SD}\text{of baseline score}\times \sqrt{(1}-r)$where r = the intraclass correlation coefficient. 3) Minimum detectable change (MDC) is calculated from SEM. MDC₉₅ is the measurement that has 95% confidence that this value is more than the measurement error. The equation is $\mathit{MDC}=\mathrm{SEM}\times 1.96\times \surd 2$. 4) MCID derived from effect size (ES) by multiplying SD of the baseline scores with 0.2, which is the accepted value for small ES.

Analysis was undertaken in STATA version 16.1 and a p value of less than 0.5 considered as statistically significant. Missing data was excluded from the analysis. Spearman’s correlation was used to determine correlation. The Wilcoxan rank-sum test, and the Wilcoxan matched-pairs signed-rank test was used to compare median scores for independent and dependent continuous data, respectively.

AVVQ

The translation process resulted in the Thai-version of the AVVQ (AVVQ-Thai).¹⁵ One hundred and ten (92%), 94 (79%), and 112 (94%) patients completed the preoperative, retest, and postoperative AVVQ-Thai, respectively. The mean AVVQ scores improved from 18.9 (standard deviation (SD) = 10.9; interquartile range (IQR) 10.1, 24.9) preoperatively to 7.8 (SD = 7.6; IQR 2, 11) at 1-month post operation (p < 0.001) with the mean difference of 11.1 (95%CI: 8.8, 13.4).

Reliability

Cronbach’s alpha of the preoperative and postoperative AVVQ were 0.783 and 0.790, demonstrating acceptable internal consistency. The overall ICC for the total scores was 0.67 (95%CI: 0.50, 0.79), demonstrating moderate reliability. Item No.8 had the lowest ICC of 0.51 (95%CI: 0.27, 0.67), while item No.3 had the highest ICC of 0.85 (95%CI: 0.79, 0.90). All items had ICCs that were more than 0.5, and 3 out of 13 (23%) had ICCs that were more than 0.75 (Table 1).

Validity

Both pre-and postoperative AVVQ were significantly correlated with pre- and postoperative VCSS, with Spearman’s correlation coefficient of 0.353 (p < 0.001) and 0.457 (p < 0.001), respectively. Median AVVQ score comparisons between groups with different degrees of problems by EQ-5D are shown in Table 2. AVVQ was able to discriminate patients with different levels of problems across all EQ-5D dimensions with significant differences in median AVVQ scores between groups in both pre-and postoperative periods except for postoperative self-care.

AVVQ scores and VCSS improved significantly after operation in all severity classifications, see Table 3. Twenty-five patients (21%) reported having no problem for all EQ-5D dimensions preoperatively, whereas no patients had the lowest possible score for the AVVQ. For these 25 patients, there were changes in AVVQ scores with the pre-and postoperative median scores of 12.2 (IQR 5.5, 21.9) and 4.2 (IQR 0, 8.2) (p = 0.002), respectively.

Minimal clinically important differences (MCID)

The MCIDs of the AVVQ varied from 2.18 to 17.16 across the four methods used (Table 4). The lowest value was from accepted small ES MCID, which was 2.8, followed by 6.17, 6.21 and 17.16 from SD_{observerd change}/2, SEM, and MDC₉₅, respectively.

Ethics and consent

Ethic committees (The Human Research Ethics Committee of Thammasat University Medicine) with approval number MTU-EC-SU-1-130/61 (approved date July 26, 2018) for Thammasat University Hospital; The Research Ethics Committee 4 of Faculty of Medicine, Chiang Mai University with approval number SUR-2562-06329 (approved date June 28, 2019) for Maharaj Nakorn Chaing Mai Hospital; The Office of The Committee for Research, Faculty of Medicine Ramathibodi Hospital Mahidol University with approval number MURA2019/173 (approved date March 13, 2019) for Ramathibodi Hospital; The Research Ethic Committee of Chonburi Hospital with approval number 26/62/S/h3 (approved date March 20, 2019) for Chonburi Hospital.

All participants signed informed consent before entering the study.