Cohort study

Sample characteristics

Primary cohort

A total of 1176 eligible people (i.e. first-time referees to PDUs during the recruitment period) were included in the analysis, 277 (23.55%) from SWLSTG, 308 (26.19%) from LPFT, 387 (32.91%) from BSMHFT and 204 (17.35%) from SHSCFT. A total of 609 participants (51.79%) were female, with only BSMHFT having a majority of male participants (216; 55.81%). Mean age of participants across sites was 36.38, while 821 (69.81%) of all participants were White British and missing data on sexuality were high (for 842, 71.60%, of participants there were no recorded data). A total of 871 participants (74.06%) had no recorded mental health diagnosis. Many people on first visit to PDU were also relatively new to mental health services, with 372 (38.27%) having had first contact with mental health services in the 9 months preceding their first PDU visit (based on data from 972 participants only, excluding participants from SHSCFT, where these data were not available). The largest single source of referrals to the PDU, across sites, was directly from the ED (488; 41.50%), with crisis and home treatment teams providing the second highest number of referrals (246; 20.92%), although these data are somewhat obscured as 157 (76.96%) of participants in SHSCFT were referred to the PDU via a single point of access to adult mental health crisis care.

In the 9-month period following discharge from the PDU, participants were most often discharged back to their usual place of residence (404, 41.56%, of 972 participants, with data on discharge destination missing from the SHSCFT site), followed by crisis and home treatment team (156; 16.05%) and acute psychiatric hospital admission (124; 12.76%). For participants across all four sites, on average, just over 50% of the follow-up period was spent while social distancing measures as a result of the COVID-19 pandemic were in place. Comprehensive data on participant characteristics by site are given in Appendix 6, Table 50.

Pre-pandemic cohort

Data on a pre-pandemic cohort of first-time referrals to PDUs in a similar period 12 months prior to our primary cohort was only available from three of the four sites as the SHSCFT PDU had yet to be opened at that point in time. A total of 934 people from the three sites were included in the pre-pandemic cohort, compared with 972 in the primary cohort, with fewer people included from SWLSTG (218 compared with 277) and LPFT (289 compared with 308) and more from BSMHFT (427 compared with 387). Characteristics of cohorts are broadly similar across all sites. Notably, the SWLSTG site accepted a much higher percentage of referrals to the PDU from CMHTs in the pre-pandemic cohort (35.3%) compared with the primary cohort (11.2%) and a correspondingly lower percentage of referrals from crisis and home treatment teams (17.9% compared with 38.6%).

Data on discharge destination were only available for the SWLSTG and LPFT sites. In SWLSTG, a slightly higher percentage of people were discharged to usual place of residence and lower to crisis and home treatment team in the pre-pandemic cohort (67.4% and 8.3%, respectively) compared with the primary cohort (51.3% and 17.0%). The discharge data from LPFT are hard to compare between cohorts, given that missing data are higher in the pre-pandemic data (24.57% compared with 9.42%), with no participants reported as discharged to usual place of residence in the pre-pandemic cohort. This may be due in part to changes in the way data are recorded. A full set of participant characteristics in the pre-pandemic cohort can been seen in Appendix 6, Table 51.

Pre and post first psychiatric decision unit visit findings by site (primary cohort)

South West London and St George’s Mental Health NHS Trust

The general picture presented by the SWLSTG data is of a marked increase in most mental health service use in the 9 months following first admission to the PDU compared with the 9 months preceding admission (see Table 18). For our primary MHT outcome, numbers of informal psychiatric admission increased highly significantly from two in the pre-admission period to 54 post admissions (χ² = 48.29, p < 0.001). However, mental health presentations at ED, our primary ED outcome (as measured by liaison psychiatry episodes in the ED) decreased highly significantly from a mean of 1.33 per participant pre admission to 0.67 post admission (t = 5.98, p < 0.001).

Most MHT secondary outcomes also demonstrated significant increases, including highly significant increases in numbers of compulsory admissions, total admissions and 0- to 5-day (short) admissions, and highly significant increases in mean number of contacts with community mental health services, including CMHTs and CRHTs. Of those community mental health service contacts, there were highly significant increases in both face-to-face and remote contacts (which nearly tripled). Increases in average length of inpatient stay (for those who had an admissions) and in mental health support line use was non-significant. There were increases in use of many other mental health service types (numbers of outpatient ED contacts fell slightly), noting also that 31 of the SWLSTG cohort had one or more subsequent additional admissions to the PDU in the 9 months following their first admission.

Lincolnshire Partnership NHS Foundation Trust

Results from LPFT reflected SWLSTG findings very closely (see Table 18). In terms of primary outcomes, there was a highly significant increase in numbers of informal psychiatric admissions from 17 in the 9 months prior to first admission to the PDU compared with 78 in the post-admission period (χ² = 52.41, p < 0.001), while there was a highly significant decrease in mental health presentations at ED from a mean of 0.68 per participant pre admission to 0.45 post admission (t = 3.30, p < 0.001).

Again, there were highly significant increases in numbers of compulsory admissions and total admissions, and significant increases in 0- to 5-day admissions from the pre to post period. There were highly significant increases in mean number of contacts with community mental health services, including CMHTs and CRHTs and, again, of those community mental health service contacts, there were highly significant increases in both face to face and remote contacts (which nearly quadrupled). There was a non-significant increase in length of inpatient stay. As with SWLSTG, there were increases in many other service use types (with a slight drop in ED outpatient appointments), and 54 people experienced one or more subsequent PDU admissions in the 9 months post first admission.

Birmingham

The picture for BSMHFT was slightly different (see Table 18). We did see a highly significant increase in informal inpatient admissions from pre to post first PDU admission – from 13 to 64 (χ² = 36.63, p < 0.001) – but no change in mean number of mental health presentations at ED (1.64 to 1.73; t = 0.98, p = 0.33). There were highly significant increases in compulsory and total inpatient admissions but no significant change in short, 0- to 5-day admissions and no significant change in average length of inpatient stay. There were highly significant increases in community mental health service use including contacts with CMHTs, and significant increases in contacts with CRHT teams. There was no change in mean numbers of face-to-face contacts but a six-fold increase in remote contacts. Atypically, numbers of contacts with a wide range of other mental health services dropped from the pre to post first PDU admission periods, and 54 people had one or more subsequent PDU admissions.

Sheffield Teaching Hospital NHS Foundation Trust

SHSCFT saw no change in either primary outcome – informal psychiatric admissions or mental health presentations at ED – from the pre-PDU first visit to post-PDU visit periods (see Table 18). There was, however, an increase in overall admissions (from 24 to 41; χ² = 6.42, p = 0.01), compulsory admissions (from 20 to 37; χ² = 7.41, p < 0.01) and a highly significant increase in average length of inpatient stay (from 19.63 to 54.61 minutes; t = 4.03, p < 0.001). Mean number of contacts with community services was significantly higher (increasing from 25.45 to 36.6; t = 3.71, p < 0.001) and there were more remote contacts (up from a mean of 9.08 to 17.17; t = 5.16, p < 0.001) while numbers of face to face contacts remained unchanged. There were more mean contacts with CMHTs (15.65–21; t = 2.56, p < 0.01), CRHT teams (8.62–13.73; t = 2.62, p < 0.01) and with the trust’s single point of access (1.18–1.88; t = 1.98, p < 0.05). As with SWLSTG and LPFT, numbers of contacts with a wide range of other community-based services increased from the pre- to post-first PDU visit periods. There were 20 subsequent PDU visits in the post first-visit period.

Pre-pandemic comparisons by site

As described in Chapter 2 for the cohort study, the same set of pre- and post-first PDU visit comparisons were made for our additional, pre-COVID cohort (see Appendix 6, Table 51), and then service use for the pre-COVID cohort, in both pre- and post-PDU visit period (see Appendix 6, Table 52), compared with our primary cohort (see Appendix 6, Table 53). These are presented below for all sites except SHSCFT, where the PDU had yet to be opened in the pre-COVID period.

South West London and St George’s Mental Health NHS Trust

There were few differences observed in either the pre- or post-PDU visit periods between the pre-pandemic and primary cohorts in SWLSTG (see Appendix 6, Tables 52 and 53). In the pre-PDU visit period there was a higher proportion of informal (8/218 people who had an informal admission vs. 2/277; OR 5.36, p < 0.05), total (14/218 with 6/277; OR 3.10, p < 0.05) and 0- to 5-day (7/218 vs. 1/277; OR 9.16, p < 0.05) admissions in the pre-pandemic cohort compared with the primary cohort, although these analyses are based on very low numbers of admissions. In the post-PDU visit period there were no differences in admissions of any kind or in average length of inpatient stay for those with an admission or in numbers of ED mental health presentations. While there was no change in mean number of community mental health service contacts from pre-pandemic to primary cohort (during the pandemic), mean numbers of face to face contacts with community services were significantly lower (13.62 vs. 17.6; t = 2.51, p = 0.01). There were also fewer CRHT team contacts in the primary cohort compared with pre-pandemic cohort (a mean of 9.18 vs. 14.83), a difference that was highly significant (t = 4.15, p < 0.001).

We also note that patterns in changes in service use from pre- to post-first PDU visit were similar across pre-pandemic and primary cohorts. In general, service use increased in both cohorts, although in the pre-pandemic period there was no increase in mean number of remote contacts with community mental health services (as was observed in the primary cohort, during the pandemic) and no increase in mean CMHT contacts.

Lincolnshire Partnership NHS Foundation Trust

Similarly, there were few differences between pre-pandemic and primary cohort in LPFT (see Appendix 6, Tables 52 and 53), including in either primary outcome (informal admissions or ED mental health presentations). In the pre-PDU visit period, there were proportionally more total admissions in the pre-pandemic cohort compared with primary cohort (24/289 people experienced one or more admissions vs. 11/309; OR 2.45, p = 0.01) and while mean numbers of face to face contacts with community services were lower (5.89 vs. 7.71; t = 2.10, p = 0.04), remote contacts higher (7.08 vs. 4.32; t = 3.21, p < 0.01). In the post-PDU visit period, there were again proportionally more total admissions in the pre-pandemic cohort compared to primary cohort (97/289 vs. 51/309) and this finding was highly significant (OR 2.55, p < 0.001). There were more contacts with community mental health services post-PDU visit in the primary cohort compared with the pre-pandemic cohort (a mean of 27.38 vs. 21.94; t = 2.36, p < 0.05) and more of those contacts were remote (a mean of 15.31 compared to 7.96), this finding being highly significant (t = 4.97, p < 0.001).

As observed in SWLSTG, patterns in pre- to post-first PDU visit changes in service use were highly similar across pre-pandemic and primary cohorts in LPFT, with significant increases in both inpatient and community service use varying only in magnitude between cohorts. The only difference observed was the significant increase in remote community contacts in the primary cohort (during the pandemic), not seen in the pre-pandemic cohort.

Birmingham and Solihull Mental Health NHS Foundation Trust

BSMHFT demonstrated even fewer differences between pre-pandemic and primary cohorts (see Appendix 6, Tables 52 and 53), with only a higher mean number of CRHT team contacts in the primary cohort (5.22 vs. 4.9; t = 6.48, p < 0.001) being different in the pre-PDU first visit period. In the post-PDU first visit period, there were no differences in inpatient service use between cohorts, while there were highly significant findings showing fewer mean face to face community contacts in the primary cohort (2.45 vs. 4.28; t = 3.62, p < 0.001), more mean remote contacts (3.06 vs. 0.69; t = 6.00, p < 0.001), and fewer mean CRHT team contacts during the pandemic (8.64 vs. 12.31; t = 50.03, p < 0.001).

Again, patterns in increased pre-post PDU visit service use were highly similar in both cohorts in BSMHFT, the only difference being an increase in face to face community contacts observed in the pre-pandemic cohort not being replicated in the primary cohort (during the pandemic).

Equalities impact assessment

There were a small number of differences in sociodemographics observed between samples in our primary cohort and a recent sample of all people using mental health services in each site (see Table 19). In all sites, the PDU sample was significantly younger that the population of people using each trust’s services as a whole. There were no other differences between groups in SWLSTG and LPFT. In BSMHFT, first-time visitors to the PDU were also less likely to be female (171/387; 44.19%) compared with the trust-wide population (37,669/69,513; 54.19%; z = 2.62, p < 0.01) and less likely to be White British (257; 66.41% vs. 41,019; 59.01%; z = 2.41, p = 0.02). In STHFT, the PDU sample was more likely to be black than the trust population (11/204, 5.39% vs. 828/31, 190 2.65%; z = 2.06, p = 0.04), although numbers were low in the PDU sample.