Pain and musculoskeletal disorders (MSDs): can an app help?

We conducted an exploratory analysis based on usage data from Andrew®.

Context

Today, the use of applications in the management of musculoskeletal disorders is generating increasing interest and has been the subject of many studies. A recent randomized controlled trial explored the impact of this type of tool, suggesting positive effects on pain and function in the intervention group compared with baseline values, and greater improvements than the control group in 67% of cases.

At Andrew®, we therefore tested this on our internal data. 

As part of this exploratory analysis, we formulated the following question:

Is use of the Andrew®App tool associated with a reduction in perceived pain in the context of musculoskeletal disorders?

The objective is to test, using usage data from Andrew®, whether a digital exercise-prescription tool is correlated with an improvement in pain related to musculoskeletal disorders.

Methodology and data

To carry out this exploratory analysis, we retrieved the data related to the health questionnaires that patients answer using Andrew®App. The inclusion criteria are as follows: 

• Patient who consulted a therapist using Andrew®App.

• Patient registered on Andrew®App.

• Patient with pain recorded by a therapist at day 0.

• Patient who completed the health questionnaire exactly at day 7.

• Patient who received an Andrew®App exercise program.

• Patient with pain at day 0 (VAS strictly greater than 0).

Pain intensity is rated using a visual analog scale (VAS) ranging from 0 to 10. The VAS is a scale commonly used in clinical practice to measure perceived pain.

The data were previously anonymized for obvious confidentiality reasons.

Regarding the statistical method, the investigations were organized as follows: 

• Descriptive statistics

• Data visualization

• Wilcoxon signed-rank test (for paired samples)

Results 

Descriptive statistics 

Sample size: 478 patients

We can observe that pain improves by an average of 1.55 between day 0 and day 7.

Data visualization

Wilcoxon signed-rank test

We know that our data do not follow a normal distribution (a Shapiro-Wilk test was performed beforehand). To assess the difference between our two samples, we therefore used a nonparametric test, the Wilcoxon signed-rank test. 

Hypotheses:

• H₀ (null hypothesis): the median of the differences between the paired values of the two samples is zero (there is no significant difference between the two samples).

• H₁ (alternative hypothesis): the median of the differences is different from zero (there is a significant difference between the two samples).

We obtained the following results: 

With a 5% significance threshold, we can reject the null hypothesis, so the difference between our two samples is statistically significant.

Interpretation 

In light of the previous results, we can therefore say that use of the tool appears to be correlated with a significant improvement in perceived pain. Associated with these results, we can put forward several hypotheses: 

• The exercises themselves have an effect on pain 

• The therapist’s management at day 0 had its effect over the following 7 days

• The patient’s self-management helps improve perceived pain 

Identified sources of bias

Selection bias

Only patients who answered the day 0 and day 7 questionnaires are included. This sample may not represent all users. Those who dropped out and did not answer the questionnaires are not taken into account.

Self-report bias

Pain at day 0 is requested from the patient by the therapist, and pain at day 7 is entered by the patient in the app. The patient may be influenced by several factors (mood, context, desire to please the therapist, pain communicated at day 0, etc…). 

Confounding bias 

We were not able to control for other external factors that could explain an improvement. (Medications, third-party interventions, etc..) 

Bias due to the absence of a control group

In the absence of a control group, we cannot conclude that the tool had a causal effect on the observed improvement.

Regression to the mean

Part of the observed improvement may be explained by a natural regression toward the mean, especially if the pain was particularly severe at the time of the first measurement.

Conclusion

This exploratory analysis suggests an improvement in perceived pain among patients using Andrew®App. These results are encouraging and should be interpreted with caution, taking into account the potential biases. This exploratory analysis was conducted with the goal of sharing knowledge gained from the Andrew®App application. We hope that it helps initiate more rigorous and controlled studies on this topic. 

Disclaimer

The results presented come from an internal analysis of anonymized data and remain exploratory results. They are not intended to replace a properly designed scientific study. This is not a clinical trial. No control group is included. The analysis cannot establish a causal link between the tool and pain reduction.

These results are therefore presented humbly, with the aim of sparking reflection and potential future studies.

Conflict of interest

The authors of this analysis work for the company that publishes the Andrew®App application. The results should be interpreted in light of this conflict of interest.