Your options depend how the data was reported. If it was reported as mean±sd for each group, then you can theoretically analyse it as continuous data with the MD/SMD as a summary measure. Your findings will tell you the estimated difference in the ordinal outcome between the groups. However, this is sometimes problematic because many ordinal scales cannot be considered to be normally distributed and therefore the mean±sd may not approximate the underlying data very well.

It is generally more appropriate to dichotomise the score at an accepted threshold (high or low), for which you can calculate an odds or risk ratio instead. Another, rarer, option is to pool the odds ratios from ordinal shift analyses if these were reported in the original trial, which is now quite common. The final option is to perform a pooled ordinal shift analysis (proportional odds regression) yourself if the number of patients in each group is available.

It generally comes down to conventions for how that scale is reported in primary literature and the data you have available.

You will find best answer in following link

https://training.cochrane.org/handbook/current/chapter-06

'Methods are available for analysing ordinal outcome data that describe effects in terms of proportional odds ratios (Agresti 1996). Suppose that there are three categories, which are ordered in terms of desirability such that 1 is the best and 3 the worst. The data could be dichotomized in two ways: either category 1 constitutes a success and categories 2 and 3 a failure; or categories 1 and 2 constitute a success and category 3 a failure. A proportional odds model assumes that there is an equal odds ratio for both dichotomies of the data. Therefore, the odds ratio calculated from the proportional odds model can be interpreted as the odds of success on the experimental intervention relative to comparator, irrespective of how the ordered categories might be divided into success or failure. Methods (specifically polychotomous logistic regression models) are available for calculating study estimates of the log odds ratio and its SE.

Methods specific to ordinal data become unwieldy (and unnecessary) when the number of categories is large. In practice, longer ordinal scales acquire properties similar to continuous outcomes, and are often analysed as such, whilst shorter ordinal scales are often made into dichotomous data by combining adjacent categories together until only two remain. The latter is especially appropriate if an established, defensible cut-point is available. However, inappropriate choice of a cut-point can induce bias, particularly if it is chosen to maximize the difference between two intervention arms in a randomized trial.

Where ordinal scales are summarized using methods for dichotomous data, one of the two sets of grouped categories is defined as the event and intervention effects are described using risk ratios, odds ratios or risk differences (see Section 6.4.1). When ordinal scales are summarized using methods for continuous data, the mean score is calculated in each group and intervention effect is expressed as a MD or SMD, or possibly a RoM (see Section 6.5.1). Difficulties will be encountered if studies have summarized their results using medians (see Section 6.5.2.5). Methods for meta-analysis of ordinal outcome data are covered in Chapter 10, Section 10.7.'

Thank you very much for explaining this.

Please see the article published in Iranian J of Public health titled "Age at Diagnosis of Breast Cancer in Iran: A Systematic Review and Meta-Analysis".

They have done a meta-analysis by Mantel and Haenszel method. Authors have mentioned in the methodology, that in case median and interquartile range of the variables were expressed, to aggregate data in a meta-analysis, Hozo et al. method was used to convert the median and interquartile range to mean and standard deviation.

Hozo SP, Djulbegovic B, Hozo I (2005). Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol, 5:13.