MHSDS#

The MHSDS contains two date columns: Diagnosis Date and Coded Diagnosis Timestamp. Only the first diagnosed date was considered. The dates were obtained using the following steps:

1. If a date was present in either column, or only in the Coded Diagnosis Timestamp column, it was retained. If the date was present in the Diagnosis Date column, it was copied into the Coded Diagnosis Timestamp column, which was then considered the primary diagnosis date.

2. Participants with missing diagnosis dates in both date columns were removed.

3. The dates and diagnosis codes were filtered accordingly.

HES#

1. HES OP records used the appointment date, and HES APC records used the admission date, which was considered to represent the first diagnosis of SMIs. No missing dates were identified for these cases.

2. Dates and diagnosis codes from HES OP and HES APC were filtered.

All health outcome-specific filtered data from each dataset were then combined. Participants were grouped using cohort key (individual - level identifiers, used to uniquely identify participants from the LPS), and by diagnosis. The earliest date for each diagnosis was considered.

Participants without any diagnosis dates who were removed earlier were checked. There were two types of missing participants as listed below:

1. For participants whose diagnoses were initially recorded with dates but later appeared as missing for the same diagnosis, the earlier dates were prioritised.

2. For participants with diagnoses recorded in the database without diagnosis dates, dates were imputed using the earliest recorded diagnosis date from a previously diagnosed condition.

The dataset was then pivoted to one row per patient, and the earliest diagnosis date for each SMI category was derived from individual ICD-10 codes. The source diagnosis column indicates the dataset from which the first date of diagnosis was recorded, specifying whether the source was MHSDS, HES OP, or HES APC. For example, in cases where a participant had received diagnoses of both paranoid schizophrenia and simple schizophrenia on different dates, and these diagnoses originated from two separate data sources (HES and MHSDS, respectively), the source diagnosis column was populated with the data source linked to the earliest diagnosis date.