Psychiatric hospital admissions and linkages to ambulatory services in the Western Cape Province of South Africa (2015-2022): trends, risk factors, the impact of the COVID-19 pandemic and possible opportunities for intervention

In recent years, hospital managers have reported increasing numbers of psychiatric hospital admissions in the Western Cape Province of South Africa, driven by poverty and substance use. We aimed to examine this trend, and the impact of the COVID-19 pandemic, as well as assess factors associated with linkage to ambulatory services following hospital discharge and repeat psychiatric admissions. Using electronic health data from the Provincial Health Data Centre, a consolidated routine service database, all psychiatric hospital admissions in the Western Cape public sector from 2015 to 2022 were analyzed, stratified by hospital level. Mixed effects logistic regression was used in this cohort study to determine the factors associated with successful linkage to ambulatory services within 30 days following hospital discharge, and repeat psychiatric admission within 30 and 90 days. We found that psychiatric hospital admissions, particularly at the district/acute level, were increasing prior to 2020 and an increasing proportion of diagnoses were substance related. 40% of admissions at the district level had not been seen at a primary health care facility in the year before the admission. Male patients and those with substance use disorders were less likely to be successfully linked to outpatient services following discharge. Successful linkage was one of the most protective factors against readmission within 90 days with an adjusted odds ratio of 0.76 (95%CI 0.73-0.79) and 0.45 (95%CI 0.42-0.49) at district/acute and specialized hospitals respectively. Improving linkage to ambulatory services by mental health patients post-discharge is likely to avert hospital readmissions.

this trend, and the impact of the COVID-19 pandemic, as well as assess factors associated with linkage to 48 ambulatory services following hospital discharge and repeat psychiatric admissions.

49
Using electronic health data from the Provincial Health Data Centre, a consolidated routine service 50 database, all psychiatric hospital admissions in the Western Cape public sector from 2015 to 2022 were 51 analyzed, stratified by hospital level. Mixed effects logistic regression was used in this cohort study to 52 determine the factors associated with successful linkage to ambulatory services within 30 days following 53 hospital discharge, and repeat psychiatric admission within 30 and 90 days. 54 We found that psychiatric hospital admissions, particularly at the district/acute level, were increasing 55 prior to 2020 and an increasing proportion of diagnoses were substance related. 40% of admissions at the 56 district level had not been seen at a primary health care facility in the year before the admission. Male 57 patients and those with substance use disorders were less likely to be successfully linked to outpatient 58 services following discharge. Successful linkage was one of the most protective factors against 59 readmission within 90 days with an adjusted odds ratio of 0.76 (95%CI 0.73-0.79) and 0.45 (95%CI 0.42-60 0.49) at district/acute and specialized hospitals respectively.

61
Improving linkage to ambulatory services by mental health patients post-discharge is likely to avert 62 hospital readmissions.

64
Manuscript word count: 3 920 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 24, 2023. ;https://doi.org/10.1101/2023 109 of Care Record), the electronic discharge summary that includes information on discharge diagnoses using . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Firstly, descriptive data looking at trends in admission numbers, diagnoses and outcomes after discharge 128 are presented. First contacts with the public health system, following hospital discharge are also explored.

129
While this is not limited to mental health ambulatory services only, as these can be difficult to distinguish 130 electronically at the primary health care (PHC) level, the assumption is that as all patients are given a 131 follow-up date after discharge, a visit directly after discharge was assumed to be most likely related to the . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

153
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 24, 2023. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 24, 2023. ; https://doi.org/10.1101/2023.05.17.23290107 doi: medRxiv preprint health data in an anonymised dataset where the researcher had no access to identifying information. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 24, 2023. ; https://doi.org/10.1101/2023.05.17.23290107 doi: medRxiv preprint Figure 2b illustrates the primary discharge diagnosis in specialized hospitals for general acute psychiatry, 200 where schizophrenia and other psychotic disorders make up more than half the primary diagnoses, with 201 mood disorders contributing to less than 20% of admissions. Substance related disorders as the primary

246
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 24, 2023. ; https://doi.org/10.1101/2023.05.17.23290107 doi: medRxiv preprint A median of 12 and 16 days were taken to first contact with the health system after discharge from 252 district/acute hospitals and specialized hospitals, respectively. For those discharged directly from an 253 district/acute hospital in 2022, 53.2% of individuals had an outpatient visit within 30 days, a number which 254 has been increasing slowly over the last few years (Figure 6a). An additional 10.43% do make contact with 255 out-patient services but take >30 days after discharge. 8.37% of discharges only make contact with the

268
Since 2015 there has been a general trend of increasing PHC visits in the period before an admission. 2020, 269 however, was the exception with the proportion of admissions with a PHC visit in the 30 days prior to 270 admission decreasing in May to July. Despite this general increase, the median days between the last PHC . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 24, 2023. in these analyses, the detailed characteristics of which are described in Table 1.

294
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

317
Sex was not associated with readmission but a primary diagnosis that was substance related had the 318 lowest association with readmission. Previous admissions were strongly associated with readmissions, as 319 well as a PHC visit in the preceding year and being a direct admission to a specialized hospital (i.e., patients . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 24, 2023. ; https://doi.org/10.1101/2023.05.17.23290107 doi: medRxiv preprint not transferred from a district/acute hospital), although the latter two factors only to a lesser degree.

321
Longer lengths of stay were also protective of readmission, although the aOR for 60-90 days and ≥ 90 days 322 similar. There was no clear trend for admission years.

323
The alternative approach of using standard logistic regression on only the latest hospital admission for an 324 individual found similar results (Supplementary Table 1  The main finding of this study is that a high proportion individuals who require psychiatric hospital 337 admission, are not routinely attending at ambulatory services, both before and after the admission, 338 suggesting that opportunities to avert psychiatric hospital admissions as well prevent repeat hospital 339 admissions are being missed. This is particularly pertinent, given the marked increase in psychiatric 340 hospital admissions that the province is experiencing, particularly at the district/acute level. Substance 341 related disorders are contributing to not only the increase in admission numbers being seen, but once an 342 individual is discharged home, those with a substance related disorder are less likely to successfully link 343 to ambulatory services. Successful linkage to care post-discharge was found to be the most protective 344 factor against repeat psychiatric admission.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 24, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 (7). While the substance rates we found were lower, they 354 were dependent on clinician ICD-10 coding, which is often not detailed enough, especially at the 355 district/acute hospital level.

356
Increased admission numbers can also be related to increased access to in-patient health services,

366
24% of mental health inpatients nationally are being readmitted to hospital within three months of being 367 discharged (14). Our study found slightly lower rates of 18-19% within 3 months, and no increase in the 368 readmission rate, except for the relative increase in district/acute hospitals in 2021, although with rising . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

404
Successful linkage to care after discharge was one of the most protective factors against readmission 405 within 30 and 90 days, at both hospital levels. All the other covariates in these analyses are relatively fixed, 406 with the health system having limited control over them, the exceptions being length of hospital stay -407 although with limited bed capacity this ability is also constrained (PHC visits in the year prior to admission 408 were associated with an increased risk of readmission, possibly indicating the chronicity of the underlying 409 disease more than the subsequent health seeking behaviour). Linkage to ambulatory psychiatric services 410 remains a longstanding challenge globally (23). And this study further highlights the importance of linkage 411 to care post-hospital discharge, which is a major challenge in the province, not limited to psychiatry only.

412
Patients diagnosed with tuberculosis in hospital, for instance, are more likely to be lost to follow up and 413 have higher mortality than those diagnosed at clinics (24). Effective clinical bridging strategies proposed 414 have included communicating discharge plans to outpatient clinicians, family involvement during the 415 hospital stay and starting outpatient programs before discharge (23).
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 24, 2023. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 24, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023