Introduction

Comorbidity is the relationship between two diagnoses. A problem with this method is that symptoms are then seen as passive indicators of hidden disorders. Comorbidity then comes from a single cause that causes the two disorders. In clinical psychology, however, it is true that symptoms are not seen as passive indicators of disorders, but that they themselves can cause other symptoms. There is a direct relationship between symptoms. There is no single cause for comorbidity, but there are a number of ways to comorbidity, and which way that is varies per person. A network approach has been developed for mental disorders and comorbidity. In this approach, symptoms are not seen as indicators of hidden disorders, but as components of a network. Comorbidity comes from a direct relationship between symptoms of multiple disorders. Half (47.4%) of all symptoms in the DSM4 are directly or indirectly connected to a main component. This component of connected symptoms has the characteristics of a "small world".

Results

Study 1: Setting up and analyzing the DSM-IV network

All symptoms from the DSM-IV were represented as nodes. There is a connection between nodes when two symptoms are both among the criteria of a particular disorder. Symptoms of the same disorder get a direct connection, while indirect connections arise when two symptoms both have a shared symptom. Two symptoms have no connection if there is no possibility of moving from one symptom to another through other symptoms.

We expect that symptom groups in the DSM are causally related to each other. Symptoms cluster more together than one can expect based on chance. Half of all symptoms therefore have a connection with each other. And the paths to get from one symptom to another are relatively short.

The degree of a node indicates how many connections the node has with other nodes. The degree distribution of a network provides important information about the network structure. The degree distribution of the DSM network is exponential. This can be seen as a "single-scale network". The symptom with the highest degree is insomnia, followed by psychomotor unrest, psychomotor disorder and depressive mood.

Another important characteristic of knots is their "betweenness". This measures the probability that a node lies between two other nodes. The four symptoms with the highest "betweenness" are irritation, distractibility, anxiety and depression.

Study 2: Empirical data for the network structure

The "small world" feature means that symptom activation is quickly spread throughout the network. The network model explains a significant part of the comorbidity within the DSM-IV. If one has one symptom from the DSM-IV, chances are that he / she will also develop another symptom from the DSM. The average shortest path length between two disorders is equal to the expected number of paths a person has to take to achieve a symptom of disorder A from a symptom of disorder B. The higher this path length is, the further two disorders lie apart in the network.

Study 3

The previous two analyzes show that path lengths and empirical comorbidity values ​​were correlated if one would expect if symptoms had a network structure. However, disorders are time dependent. They are dynamic, and sometimes there is a time criterion for how long the symptoms should be present as a minimum. The network model takes this into account. To show how the network can predict comorbidity, a simulation was performed for two disorders to show that the comorbidity values ​​are consistent with empirical data. Symptom dynamic is an increase in the chance of a certain symptom, when someone has more neighbor symptoms. If someone has depression and a loss of interest, the chance of having suicidal thoughts is higher than for someone who has a loss of interest, but no depression. Even though the simulated networks are incomplete, the results do provide evidence for the possibility that a network model can reproduce empirical data.

Discussion

The "missing heritability" indicates that the individual differences in vulnerability to develop a disorder are for the most part genetically determined. However, only a small part of the genetic variance can be traced to identified gene changes.

The strength of the symptom connections in a network can partly be explained genetically, but it is probably not the case that all connections in a disorder are influenced by the same genes in all people.