##### Authors: Sharon Yeo1, Han Tun Aung2 and Louis Tong1,3,4,5*
Affiliations:
1Singapore Eye Research Institute, Singapore
2School of Health Sciences, Ngee Ann Polytechnic, Singapore
3National Eye Centre, Singapore
5Yong Loo Lin School of Medicine, National University of Singapore
Dates:
Received: 19 May, 2014; Accepted: 27 June, 2014; Published: 30 June, 2014
*Corresponding author:
Louis Tong, Singapore National Eye Center, 11 Third Hospital Avenue, Singapore 168751, Tel: +65-62277255; Fax: +65-63224599; Email: Louis.tong.h.t@snec.com.sg
Citation:
Yeo S, Aung HT, Tong L (2014) The Association of Dry Eye Symptoms with Socioeconomic Factors and quality of Life. J Clin Res Ophthalmol 1(1): 006-013. DOI: 10.17352/2455-1414.000002
© 2014 Yeo S, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Keywords:
Dry eye; Quality of life; Dry eye symptoms

Purpose: Dry eye is a common condition with significant morbidity and socioeconomic burden. The associated demographic factors that worsen utility in dry eye patients were not known. There were many questionnaire instruments advocated for dry eye documentation but none of these have been shown to correlate to quality of life (QoL). We aimed at examining the health related utility values in a group of dry eye patients and their associations.

Methods: This was a hospital based prospective cross-sectional study conducted at the dry eye clinic of Singapore National Eye Centre. Patients with dry eye symptoms were randomized to one of the two validated symptom questionnaires, Standard Patient Evaluation of Eye Dryness (SPEED) or Symptom Assessment in Dry Eye (SANDE) questionnaires. All patients underwent an evaluation of socio economic factors and utility was assessed using Time-Trade-Off methodby 4 trained interviewers.

Results: We recruited 178 participants with dry eye symptoms (mean age was 56.4 (SD: 14.1) years, 77% female), 85 were assessed with SPEED and 93 with SANDE. The utility values encountered were skewed with only 52% of patients having a reduced utility (median=1.0) with a mean of 0.984 (SD: 0.11). The mean SPEED was 11.7 (5.6), and the mean SANDE was 56.8 (22.6). A higher symptom score was associated with a utility less than 1 with odds ratio 2.75 (95%CI 1.50-5.04). The correlation between SANDE and utility was r=-0.295 (-0.47 to -0.097) and corresponding correlation for SPEED and utility was not significant. To detect a utility less than 1, SPEED had an area under the curve (AUC) of 0.63 (95%CI 0.51-0.75) and the SANDE, 0.67 (0.56-0.77).

Conclusions: The health related QoL was relatively good in people with dry eye symptoms in this study. Increased symptoms were associated with decreasing QoL but the association was in general weak. This implied that causes of reduced QoL apart from symptoms, such as costs, treatment inconvenience or adverse effects should be explored.

### Introduction

1. ##### Table 4:

Mean and standard deviation in symptoms score for participants with utility of less than 1 and utility of one.
†SPEED scores were transformed to 0-1 from its original scale of 0-32 before computation in this column *P values are significant at 95% CI (2 tailed test)

The inverse correlation between utility value and the dry eye symptoms is shown in Table 5. The participants with higher symptoms were correlated to lower QoL (lower utility). In general, the strength of correlation between the utility and the extent of dry eye symptoms was relatively weak, even though statistically significant in a few instances. The correlation between utility and SPEED symptom was -0.198 (-0.394 to 0.016) whereas that between utility and SANDE was -0.295 (-0.470 to -0.097). As in Table 4, the inverse correlation was significant for participants who were Chinese, female, aged 51 and above, earning lesser than S$3000 and with up to A levels of highest education, and those living in apartments with 4 rooms or smaller. An interesting finding was the moderately inversely correlated findings between utility and SANDE for participants living in private apartments (r=-0.52), in 4 rooms or smaller (r=-0.419) and those with income lesser than S$3000 (r=-0.432). There were also weak correlated findings between utility value and SPEED for female participants (r=-0.298) and those with education up to post-secondary level (r=0.282).

1. ##### Table 5:

Correlation of utility and symptoms score.
†SPEED scores were transformed to 0-1 from its original scale of 0-32 before computation in this column *P values are significant at 95% CI (2 tailed test)

Logistic regression was performed with a lowered utility as the outcome (defined as utility value of less than 1) and the symptom score (dichotomized to either above or below the mean) was used as the independent variable. A higher symptom score was associated with reduced utility, with a crude odds ratio (OR) 2.75 (95% CI 1.50-5.04). The odds ratio adjusted for age and gender was the same at 2.70 (1.46-5.00). When adjusted for age, gender, education level, income level and housing status, the adjusted OR was 2.72 (1.44-5.15).

Receiver operating curves are shown for the detection of utility of less than 1 (Figure 2). The SPEED and SANDE questionnaires had areas under the curve (AUC) of 0.63 (95%CI 0.51-0.75) and 0.67 (0.56-0.77) respectively. The most suitable SPEED threshold for predicting a decrease in utility was 11.5 out of a total of 32, with a sensitivity of 57% and specificity of 65%. However, to achieve a higher sensitivity, a lower threshold of 8.5 produced a sensitivity of 71%, but this would lower the specificity to only 42%. For the SANDE a threshold of 52 out of 100 produced sensitivity of 66% and specificity of 56%. To attain a higher sensitivity, a lower threshold of 49.3 produced a sensitivity of 78%, but specificity of only 51%.

1. ##### Figure 2:
Receiver operating curves for SANDE (left) and SPEED (right) to detect a utility of less than 1.

We repeated the above analyses without including the non-Chinese participants, and essentially the same conclusions were drawn (data not shown).

### Discussion

This study found that health related QoL represented by the utility value was relatively high in dry eye patients in the hospital-based Singapore study, with only 51.7% of patients having a utility value of less than 1 (no reduction in QoL). We found that reduction of utility value was associated with younger age of less than or equal to 50 years. In general, socioeconomic and demographic factors did not impact symptoms of dry eye. A reduced utility was associated with more symptoms of dry eye but the strength of the association was weak. Interestingly, the correlation was moderately strong when symptoms were measured with the SANDE for those living in private apartments, those living in 4 rooms or smaller apartment or those earning less than $3000. With worse dry eye symptoms, there was a 3-fold increase in risk for utility to be reduced. The SPEED threshold of 11.5 or SANDE threshold of 52 may be suitable for screening to detect a lowered utility value in dry eye. Utility valuein this study seems much better than Schiff man's study (where even mild dry eye was associated with a mean utility value as low as 0.81). The utility measurement was known to depend on an individual's aversion to risk [4]. It might be that subjects recruited for our study were less willing to take risks, or entertain the concept of death, compared to the participants in Schiff man's study. This may be attributed to cultural or religious differences. As far as we know, the current study is the first to find a correlation between the magnitude of symptoms and QoL. The relatively weak correlation between symptoms and QoL could be due to other problems faced by people with dry eye that are unrelated to the symptoms. These include socioeconomic burden of dry eye, inconvenience of dry eye treatment and follow up visits at the clinic. Between 2008 and 2009, the total expenditure of dry eye treatment in our center was found to be US$0.06 million per year per 1,000 patients. However, indirect costs such as loss of productivity were not yet estimated [5]. Dry eye treatment is inconvenient, often requiring instillation of numerous drops a day, with consequently an impact on daily activities. Some patients who were loss to follow up also reported problems with hospital visits (unpublished data). Psychological effects of dry eye were explored previously and it was found that anxiety and depression were correlated with dry eye syndromes [10].

We found a weak correlation between utility and the symptoms of dry eye in participants without systemic disease (as defined in this study), but we failed to elicit any correlation in those with systemic disease. Perhaps people with systemic conditions have decrease in QoL mainly due to those conditions and would not trade-off more time addressing the relatively less impactful dry eye condition. The reason why SANDE scores were more correlated to utility than SPEED could be due to SANDE having continuous variable readings which were more precise than the interval/ordinal measurements in SPEED.

In some scenarios, SANDE may be preferable to SPEED for assessment of dry eye symptoms, especially if the research involved possible changes in symptoms which may have an impact on QoL. However, for screening purposes, SPEED (at a threshold of 11.5), or SANDE (with threshold of 52) may be suitable as a one-time tool to detect a reduced utility.

A limitation of this study is that visual function disturbances often affecting a dry eye patient were not captured. These include forced visual acuity, contrast sensitivity, and non-invasive tear break up time, etc. Another limitation of the study is that we did not evaluate the number of hours on computer use, or attempt to classify the patients based on occupations. These activities may independently affect QoL in addition to the socioeconomics.The study has very few Sjogren syndrome patients, so the results pertaining to QoL may only apply to non-Sjogren syndrome type of dry eye patients.

### Conclusion

In conclusion, the health related QoL was relatively high in people with dry eye symptoms in this hospital-based study in Singapore. This implied that apart from symptoms alone, other factors such as costs, treatment inconvenience or adverse effects should be explored as factors that reduce QoL in these patients. With increased magnitude of dry eye symptoms, there is a 3-fold increase in the odds for utility to be reduced. The SPEED threshold of 11.5 or SANDE threshold of 52 may be suitable for screening to detect a lowered utility in dry eye.

1. ##### Supplementary Table 1:

Mean and standard deviation of utility values.
The following method of transformation was designed to ensure that the mean of transformed SPEED has the same mean as the SANDE score. For SPEED scores greater than its mean, the following formula is used:
$TransformedSPEED\text{​}score=meanSANDE+\left(100-meanSANDE\right)×\frac{SPEED\text{​}\text{​}score-meanSPEED}{32-meanSPEED}$
Else, the following formula is used: $TransformedSPEED\text{​}score=meanSANDE×\frac{SPEEDscore}{meanSPEED}$