Brief admission by self-referral as an add-on to usual care for individuals with self-harm at risk of suicide: cost-effectiveness and 4-year health-economic consequences after a Swedish randomized controlled trial

Figures & data

Figure 1. Chart with participant flow during and after the randomized clinical trial (RCT). †1 died by suicide, 3 felt too unstable and 2 were unavailable to schedule time‡From the former BA group in the RCT. §From the former control group in the RCT.

Table 1. Unit prices and cost calculation of hospital care in euros, 2020.

	Unit price (€), base case	Unit price (€), capacity-adjusted^a	Source (2020)	Assumption
Brief admission by self-referral, per patient-day	1032 (49% bed occupancy)	613 (85% bed occupancy)	Psychiatry, Lund; BA unit	Average cost per patient-day = total cost divided by number of patient-days delivered.
General psychiatric admission, per patient-day	714 (115% bed occupancy)	828 (85% bed occupancy)	Psychiatry, Lund: psychiatric emergency care unit. Statistics on surveillance^b	Average cost per patient-day = total cost divided by number of patient-days delivered. Adjusted for staff costs for surveillance (2 and 5 h per 24 h during general and compulsory admission).
Compulsory psychiatric admission, per patient-day	807 (115% bed occupancy)	921 (85% bed occupancy)
Inpatient care at other clinics, per patient-day	652–2029	652–2029	Regional price list	Cost per day according to price list^c
Visit to a psychiatric emergency department, per visit	723	723	Psychiatry, Lund	Average cost per visit = total cost 2020 divided by the number of visits in 2020
Visit to other type of emergency department, per visit	378	378	Regional list	Cost per visit according to price list^c
Compulsory measure; Belting	65 and then 30 per hour	65 and then 30 per hour	Psychiatry, Lund, and list price pharmacy^d	Cost per measure = Hourly staff wage rate × estimated time + cost per medical injection
Compulsory measure; Injection	68	68
Compulsory measure; Shielding	30 per hour	30 per hour

^aCapacity-adjusted to an annual average bed occupancy rate of 85%. ^bHours of surveillance were not reliably documented in patient journals. Statistics on hours of surveillance per type of admission were therefore provided from staff at a representative department for two separate weeks: week 49 in 2020 and week 15 in 2021. The cost per hour of surveillance was estimated to be the hourly cost for one nurse’s aid. ^cTreatment at other clinics, included medical, surgical, orthopedic, ear/nose/throat, women’s clinic, and intensive care clinic. Estimates for psychiatric care prices did not include overhead cost of regional leadership, so prices collected from regional price lists were adjusted accordingly (reduced with 1.5%), after consultation with a representative of the healthcare authority in Southern Sweden. ^dEstimated staff time per compulsory measure was reached after discussion between the last author, who has extensive experience of working with the target group; the chief physician at a representative psychiatric department; the chief physician at the psychiatric intensive care unit responsible for compulsory measures; and an additional on-call physician at Lund University Hospital. Pricing for staff time utilized the average wage rate for needed staff based on data from the Lund Hospital finance department. The price for medical injection was exemplified with lorazepam 4 mg/ml according to the pharmacy price list.

Table 2. Estimated treatment cost before and during randomized controlled trial (in euros based on 2020 prices; n = 117).

	Baseline, 1 year, (no one could access brief admission by self-referral). Referred to as year 0 (Y0)					With-trial-period, 1 year, (randomized groups to brief admission by self-referral as add on to usual care or usual care alone). Referred to as year 1 (Y1)
		Base case		Capacity-adjusted			Base case		Capacity-adjusted
		Mean	SD	Mean	SD		Mean	SD	Mean	SD
BA group	H1	29 400	37 300	33 500	42 600	H1	38 200	39 300	38 100	41 600
	H2	45 500	38 900	49 200	42 700	H2	26 100	35 100	26 000	39 200
	Y0	74 900	··	82 600	··	Y1	64 300	··	64 200	··
Control group	H1	27 700	39 100	31 700	44 900	H1	29 400	36 300	33 600	41 600
	H2	37 700	30 100	39 200	34 800	H2	20 800	29 900	23 700	34 200
	Y0	65 400	··	70 900	··	Y1	50 200	··	57 300	··
Difference		··	··	··	··		4 600⸸	··	−4 800⸸	··

BA group: randomized to brief admission by self-referral as add on to usual care during the trial year. Control group: randomized to usual care during the trial year. SD = standard deviation. H1 = first half-year. H2 = second half-year. Y = year. ⸸Difference in costs was calculated by comparing the groups in terms of the difference in costs between the within-trial period (Y1) and baseline (Y0).

Table 3. Quality-adjusted life years (QALYs) during 1-year randomized controlled trial (RCT).

	Period	Mean	SD
BA group	H1	0·364	0·045
	H2	0·375	0·045
	Y1	0·739	··
Control group	H1	0·342	0·042
	H2	0·319	0·116
	Y1	0·661	··
Difference		0·078	··

BA group: randomized to brief admission by self-referral as add on to usual care during the trial year. Control group: randomized to usual care during the trial year. SD = standard deviation. H1 = first half-year. H2 = second half-year. Y = year. The number of participants varied from 41 to 99 based on available data at each time point, see details in supplementary material, supplement 4, for information on utility after five years including three percent discount rate to account for differential timing.

Table 4. Cost-effectiveness of Brief admission by self-referral as add on to usual care compared to usual care alone (2020 prices in euros).

	Base case	Capacity-adjusted
Cost per QALY gained§ (euros), all (n = 117)	59 000	BA is cost-saving
Cost per QALY gained§ (euros), complete cases (n = 37)	BA is cost-saving	BA is cost-saving
Cost per QALY gained§ (euros), cases with <180 hospital admission days during baseline year (Y0) (n = 101)	119 900	16 500

§Cost per quality-adjusted life year (QALY) gained = (Difference in costs between the randomized groups)/(Difference in QALYs between the randomized groups).

Figure 2. Scatterplots of bootstrap re-sampling based on actual bed occupancy in 2020 (blue) and capacity-adjusted to an average of 85% bed occupancy (orange) for complete cases (n = 37, boxes a and C) and all cases, using multiple imputation on utility where missing (n = 117, boxes B and D). The horizontal axes show QALY gain with BA versus TAU. The vertical axes show incremental cost in euros (mean cost with BA versus TAU).

Figure 3. Number of days admitted to hospital (boxes a and B) and treatment costs in euros (boxes C and D), average per person per year before, during, and after the randomized controlled trial (RCT). BA: intervention group during RCT (n = 56); Control: control group during RCT (n = 61). Base case: prices based on actual bed occupancy rate; Capacity-adjusted scenario: prices based on an average bed occupancy rate of 85%. See supplement 7 for costs during follow-up, including costs discounted with 3% to account for differential timing in accordance with recommendations [25].

Attema AE, Brouwer WBF, Claxton K. Discounting in Economic Evaluations. Pharmacoeconomics. 2018;36(7):745–758. doi:10.1007/s40273-018-0672-z.

 PubMed Web of Science ®Google Scholar

Data availability statement

The clinical data supporting the study findings are not publicly available. In the approval from the regional ethical review board (2014/570), it is stated that data should only be presented on a group level as these might contain sensitive information.