Using Routinely Collected Data From the Network of Alcohol and Other Drugs Agencies Database to Evaluate the Impact of COVID‐19 Lockdowns on Trends in Service Delivery in Outpatient Non‐Government Alcohol and Other Drug Treatment Services
1 School of Psychology, Faculty of Arts, Social Sciences and Humanities University of Wollongong Wollongong New South Wales Australia
2 The Statistical Consulting Centre, National Institute for Applied Statistics Research Australia, School of Mathematics and Applied Statistics University of Wollongong Wollongong New South Wales Australia
3 Centre for Alcohol Policy Research, School of Psychology and Public Health La Trobe University Melbourne Victoria Australia
4 Network of Alcohol and Other Drugs Agencies Sydney New South Wales Australia
5 Drug Policy Modelling Program, Social Policy Research Centre University of NSW Sydney New South Wales Australia
6 National Centre for Youth Substance Use Research, School of Psychology, the University of Queensland St Lucia Queensland Australia
7 National Drug and Alcohol Research Centre University of NSW Randwick New South Wales Australia
* Correspondence:Alison K. Beck (alisonbe@uow.edu.au)
ABSTRACT
Background
The COVID‐19 pandemic disrupted face‐to‐face alcohol and other drug (AOD) treatment services and prompted greater use of telehealth. This study measured the impact of COVID‐19 on outpatient treatment episodes before, during, and after stay‐at‐home orders were introduced.
Methods
This observational study used routinely collected data from community‐based non‐government AOD services in New South Wales (NSW), Australia across 4 years (209 7‐day intervals from January 1, 2019–January 2, 2023). Interrupted time‐series analyses with seasonal autoregressive integrated moving average (ARIMA) modeling estimated weekly changes in outpatient treatment episodes associated with stay‐at‐home (lockdown) orders, including (1) commencements, (2) planned cessations, and (3) unplanned cessations. Episode counts were also examined by gender (male, female), age group (< 25 years, 25–59 years, ≥ 60 years), principal drug of concern (alcohol, amphetamines, cannabinoids, opioids, and other), and location (metropolitan and non‐metropolitan).
Results
There were no significant level or trend changes in the overall number of episode commencements (ARIMA (2,0,0)(1,0,0), lockdown one: β_level = −814.85, p = 0.098; β_slope = 11.14, p = 0.127; lockdown two: β_level = 317.3, p = 0.427; β_slope = −3.47, p = 0.237), planned cessations (ARIMA (1,0,0)(0,0,1), lockdown one: β_level = −142.1, p = 0.66; β_slope = 1.16, p = 0.808; lockdown two: β_level = −199.07, p = 0.432; β_slope = 0.18, p = 0.932), or unplanned cessations (ARIMA (1,0,0), lockdown one: β_level = −92.4, p = 0.717; β_slope = 1.4, p = 0.711; lockdown two: β_level = −121.19, p = 0.563; β_slope = 0.53, p = 0.732) in outpatient non‐government AOD services at either lockdown. For subgroups, during lockdown one, commencements per week increased for metropolitan participants (ARIMA (1,0,0) β_level = −725.12, p = 0.024; β_slope = 10.07, p = 0.035) and those with amphetamines (ARIMA(3,0,0)(1,0,0) β_level = −338.44, p = 0.036; β_slope = 4.68, p = 0.05) as their principal drug of concern.
Conclusions
The introduction of COVID‐19 lockdown measures did not appear to adversely impact the number of NSW non‐government AOD outpatient treatment episodes delivered nor the likelihood of unplanned dropout. Our findings illustrate the likely role of telehealth in sustaining (and for some groups, perhaps temporarily increasing) AOD service provision during the COVID‐19 pandemic. This highlights the sector's resilience in sustaining care under challenging circumstances.
Graphical
This observational study used routinely collected data from community‐based, non‐government alcohol, and other drug services in New South Wales, Australia. Lockdown measures did not adversely impact the number of treatment episodes that commenced or increased the likelihood of unplanned dropout. These findings highlight the sector's resilience and capacity to sustain care through rapid service adaptation.
Boxed Text
1Introduction
Alcohol and other drug (AOD) use are leading contributors to the global burden of disease (Degenhardt et al. 2018). The harms arising from AOD use are compounded by the discrepancy between treatment demand and availability, with conservative estimates of between 30% and 48% of people with an AOD use disorder going untreated (Ritter and O'Reilly 2025). The COVID‐19 pandemic posed further challenges to service utilization, with both increased use of AOD (Schmidt et al. 2021; Roberts et al. 2021) and widespread disruption to service provision (World Health Organisation 2020; Kumar et al. 2022). To mitigate the spread of COVID‐19, “lockdowns” were a key public health response, and involved restrictions around people's movement, gatherings, and the operation of businesses (Holley et al. 2021).
In Australia, the first case of COVID‐19 was diagnosed on January 25, 2020 (Department of Health and Aged Care 2020). Restrictions were introduced by federal and state governments to curtail transmission (Roth 2020; Holley et al. 2021). The level of restrictions adopted by each of Australia's States and Territories was based on directions from state‐based health authorities in response to local case numbers and hospitalization. Due to escalating COVID‐19 cases, the state of New South Wales (NSW) experienced three periods of stay‐at‐home orders (“lockdowns”; Roth 2020). The initial state‐wide lockdown was introduced on March 31, 2020, and lasted until May 14, 2020 (45 days) (Roth 2020). A local lockdown was applied to the Northern Beaches region of Sydney on December 19, 2020 (Minister for Health and Medical Research 2020), and lifted on January 9, 2021 (21 days) (Nyinawingeri et al. 2023). Lockdown measures were re‐introduced to Sydney and the surrounding local government areas from June 25, 2021, with restrictions ultimately lifted on October 11, 2021 (108 days) (Minister for Health and Medical Research 2021). The spread of COVID‐19 and the introduction of lockdowns impacted the delivery of AOD treatment services (Dunlop et al. 2020; de Vargas et al. 2021). This included a rapid transition from face‐to‐face service provision to telehealth (Lin et al. 2023).
Telehealth is the use of digital information and communication technologies (e.g., telephone and video call) to remotely provide health information, prevention, monitoring, or medical care (Doraiswamy et al. 2020). Telehealth may improve access to care, for example, by increasing accessibility (e.g., eliminating transport barriers and increasing convenience; Substance Abuse and Mental Health Services Administration 2022) and reducing stigma associated with face‐to‐face help‐seeking (Fletcher et al. 2018). A range of barriers to telehealth have also been identified, including limited digital resources and infrastructure, technological difficulties, and a preference for face‐to‐face service provision (Mark et al. 2021; Beck et al. 2023; Lin et al. 2023; World Health Organisation 2020). Disparity in the use of technology for delivering health care in low‐income countries has also been noted (World Health Organisation 2020). A recent systematic review demonstrated that utilization of telehealth during the pandemic was consistently lower among vulnerable populations, including older people and people in rural locations (Vakkalanka et al. 2024). Gender disparities were also observed (Vakkalanka et al. 2024). However, telehealth may not always exacerbate disparities for vulnerable populations (Palzes et al. 2022; Busch et al. 2023).
Findings on the impacts of the COVID‐19 pandemic on AOD service utilization globally are similarly mixed. For example, during the first year of the pandemic, medical record data points to an increase in any utilization of AOD services in Northern California (Palzes et al. 2022). In contrast, the Veterans Health administration noted an overall decline in service provision (Perumalswami et al. 2023). Preliminary data from four US‐based hospitals early in the pandemic demonstrated an initial reduction in the number of addiction consults delivered, with a subsequent return to baseline levels (Murphy et al. 2021). Similarly, longitudinal data of outpatient visits in the US point to a time‐limited reduction in the number of outpatient visits early in the pandemic (Busch et al. 2023). Conversely, evidence from Massachusetts found a time‐limited increase in outpatient AOD services, with a subsequent reduction compared with pre‐pandemic levels (Yang et al. 2020). Findings from Australia showed an increase in utilization of mental health services more broadly (Australian Institute of Health and Welfare 2021), but an overall decline in non‐government AOD services in NSW specifically (Van De Ven et al. 2021). Another Australian study found a reduction in the number of face‐to‐face episodes of counseling provided by a large provider of non‐government AOD services, but an increase in telephone‐delivered counseling (Carlyle et al. 2023). With the exception of this recent Australian study (Carlyle et al. 2023), these findings are constrained to AOD service provision during the first year of the pandemic.
To help clarify the impact of COVID‐19 lockdowns on service provision, longer term data are needed and analysis methods that account for pre‐existing trends in AOD service utilization required. As engagement with telehealth may also differ according to individual characteristics (e.g., age, gender and remoteness (Vakkalanka et al. 2024) and substance type (Tilhou et al. 2024)), subgroup analyses are warranted. Understanding these trends in commencement and cessation of AOD treatment episodes can help to inform funding bodies and service providers about the uptake of AOD services during periods of disruption. This may be important for future planning and decision‐making surrounding service provision in times of need (e.g., pandemics and natural disasters).
The objective of this study was to use interrupted time‐series analysis to investigate the effect of the lockdown measures during COVID‐19 on service delivery among a state‐wide sample of outpatient non‐government AOD services in NSW Australia. Specifically, we compared shifts in the number of treatment episode commencements and cessations (planned and unplanned) before, during, and after the introduction of COVID‐19 lockdown measures. We hypothesized that compared with the pre‐lockdown period, lockdown measures would lead to (a) a reduction in the number of commencements and planned cessations, and (b) an increase in the number of unplanned cessations (due to reduced mobility and individual/organizational challenges transitioning to telehealth) during the lockdown periods. We also explored whether the impact of lockdown measures on episode commencements and cessations differed by gender, age, principal drug of concern, and participant location.
2Method
2.1Study Design
This observational study used routinely collected data captured by the Network of Alcohol and other Drugs Agencies (NADA) database (NADAbase). This study was approved by the University of Wollongong Social Sciences Human Research Ethics Committee (2012/207). Findings are reported according to the RECORD statement (Benchimol et al. 2015).
2.2Data Source
NADAbase is an online client data repository system established by NADA for the purpose of collecting and reporting data related to quality improvement, performance monitoring, evaluation, and research. NADA is a peak body for non‐government AOD services in NSW and represents 85 publicly funded non‐government AOD members with 100 services in multiple locations. Of these, 79 agencies input their service data into NADAbase. The AOD data in NADAbase include all clients attending NSW non‐government providers of AOD services who have agreed to share data via a signed NADAbase data user agreement. NADAbase data are stored in a Microsoft SQL Server database secured with authentication controls.
2.3Setting
In Australia, non‐government organizations are a leading provider of AOD treatment. Nationally in 2022–2023, non‐government organizations provided 73% of all treatment episodes (Australian Institute of Health and Welfare 2024). The proportion of non‐government AOD treatment services in NSW is 40.7% (Australian Institute of Health and Welfare 2024). Since 2000, NADAbase has been used by member services to enter a set of standard data elements (minimum dataset) relating to an episode of care for each person who engages with the service. The routinely collected information includes the clients' demographics, treatment provision, and treatment outcomes measures. These are described in full in the NADAbase Data Dictionary (The NADA 2025).
2.4Participants
The study population were people aged ≥ 10 years who were recorded in the NADAbase as having attended an outpatient AOD service in NSW for their own AOD use during the 209 weeks between January 1, 2019, and January 2, 2023.
2.5Variables
NADAbase variables used in this analysis were date of commencement of service episode, date of cessation of service episode, reason for cessation (planned cessation = service completed; unplanned cessation = left without notice, left against advice, left involuntarily—non‐compliance), service delivery setting (outpatient), client type (own drug use), gender, age (at commencement of the service episode), principal drug of concern, and postcode (of client's usual place of residence at the commencement of the service episode). These variables are derived from the Alcohol and Other Drug Treatment Services National Minimum Data set.
2.6Observation Periods
The unit of observation for this time‐series analysis was the weekly number of each outcome (treatment episodes commenced; treatment episodes ceased: planned/unplanned) across community‐based NGO AOD services. Week number was defined as each consecutive 7‐day period beginning January 1—January 7, 2019 (week 1) to January 2, 2023 (week 209).
To have sufficient power to estimate regression coefficients, at least eight time points before and after the interruption are recommended (Penfold and Zhang 2013). Given the short 3‐week timeframe of the localized Northern Beaches lockdown (Nyinawingeri et al. 2023), we were not able to assess the impact of this interruption. As such, this period was considered as part of the post‐lockdown period that occurred after the initial restrictions were lifted. Interruptions were defined at the weeks corresponding to the start (Week 64) and end (Week 71) of the first state‐wide lockdown (Lockdown 1) and the start (Week 130) and end (Week 144) of the second state‐wide lockdown (Lockdown 2). See Appendix 1 for a timeline of lockdown restrictions and analysis observation periods.
2.7Outcomes
Using eligible data during the 209 weeks between January 1, 2019, and January 2, 2023, we examined outpatient treatment episodes, including (1) commencements, (2) planned cessations, and (3) unplanned cessations. Episode counts were also examined by gender (male and female), age group (< 25 years, 25–59 years, and ≥ 60 years), principal drug of concern (alcohol, amphetamines, cannabinoids, opioids, and other), and location (metropolitan and non‐metropolitan).
2.8Analysis
Due to typographical errors in the data set (e.g., date of birth being listed the same as date of commencement; 1900 being listed as a birth year due to data capture error), age was labeled as “missing” for values lower than 10 (based on entry criteria for the services represented in the data set) and older than 100. Gender was restricted to male and female, due to the low number of gender diverse people represented in the data set. Principal drug of concern was condensed to the four most common substances (alcohol, amphetamines, cannabinoids, and opioids) and the remainder collapsed into an “other” category. Residential postcode was used to classify people according to the five levels of “remoteness” defined by the Australian Standard Geographical Classification (Australian Bureau of Statistics 2011). Consistent with published recommendations for examining “rural and remote” Australians (Australian Institute of Health and Welfare 2023), response categories were collapsed for analysis (Major city vs. Inner Regional/Outer Regional/Remote/Very Remote).
For each week of the study period, we calculated the total number of treatment episodes that commenced or ceased (planned and unplanned) in outpatient AOD services (overall and per subgroup). The total number of episodes that commenced and ceased (planned and unplanned), overall and for each subgroup of interest, was considered as separate interrupted time‐series data. Trends were plotted to visualize changes in weekly episode counts.
To estimate underlying trends in each interrupted time series, segmented linear regression analysis was performed with seasonal ARIMA modeling of the resulting residual structure. Slope and intercept terms were introduced to estimate changes in the trend of weekly episode counts between each period of interruption. Optimal selection of ARIMA and seasonal parameters (with 52‐week lag) was performed using the fable package in R (O'Hara‐Wild et al., n.d.) which implements automatic ARIMA parameter selection (Hyndman and Khandakar 2008). Autoregressive parameters were restricted to one to five lags for model simplicity. An additional covariate which indicated whether the weekly counts occurred during the Christmas and New Year period (present if the corresponding week contained either Christmas or New Year) was introduced to account for the expected disruption in services. Assumption checking is described in Appendix 2. The analysis was not pre‐registered.
3Results
Descriptives for the number of episode commencements and cessations (planned and unplanned) overall and per subgroup for each of the study periods are presented in Table 1. Figure 1 presents the visualizations of the time‐series data. Tables 2, 3, 4 present the findings from the interrupted time‐series models.
| Pre‐lockdowns | Lockdown one | Between lockdowns | Lockdown two | Post‐lockdowns | Total | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Total | M (SD) | Range | Total | M (SD) | Range | Total | M (SD) | Range | Total | M (SD) | Range | Total | M (SD) | Range | Total | M (SD) | Range | |
| Episode commencement | ||||||||||||||||||
| Overall | 20,800 | 330.15 (77.75) | 40–478 | 2507 | 313.37 (54.73) | 196–364 | 19,255 | 331.98 (68.06) | 74–478 | 4798 | 319.86 (27.02) | 258–371 | 18,136 | 279.01 (66.98) | 13–397 | 65,496 | 313.37 (71.80) | 13–478 |
| Gender | ||||||||||||||||||
| Male | 13,362 | 212.09 (48.44) | 24–303 | 1640 | 205.00 (43.10) | 115–252 | 12,218 | 210.65 (43.12) | 35–275 | 2988 | 199.20 (19.70) | 157–235 | 11,019 | 169.52 (42.47) | 5–253 | 41,227 | 197.25 (47.04) | 5–303 |
| Female | 7358 | 116.79 (32.31) | 16–209 | 862 | 107.75 (14.14) | 81–121 | 6948 | 119.79 (30.15) | 26–205 | 1785 | 119.00 (11.92) | 97–135 | 6740 | 103.69 (25.50) | 6–141 | 23,693 | 113.36 (28.72) | 6–209 |
| Age Group (years) | ||||||||||||||||||
| < 25 | 7054 | 111.96 (34.99) | 11–194 | 797 | 99.62 (17.02) | 77–135 | 6809 | 117.39 (35.00) | 22–242 | 1457 | 97.13 (16.38) | 71–135 | 4892 | 75.26 (30.71) | 2–150 | 21,009 | 100.52 (36.58) | 2–242 |
| 25–59 | 13,024 | 206.73 (44.33) | 26–283 | 1601 | 200.12 (42.22) | 114–249 | 11,686 | 201.48 (39.57) | 39–273 | 3170 | 211.33 (13.94) | 183–235 | 12,470 | 292.84 (46.42) | 12–273 | 41,951 | 200.72 (42.37) | 12–283 |
| ≥ 60 | 639 | 10.14 (4.39) | 2–22 | 101 | 12.625 (5.42) | 2–20 | 676 | 11.65 (4.29) | 3–20 | 145 | 9.66 (3.75) | 2–15 | 779 | 11.98 (4.95) | 0–23 | 2340 | 11.19 (4.60) | 0–23 |
| Principal drug of concern | ||||||||||||||||||
| Alcohol | 6523 | 103.53 (24.60) | 13–152 | 6674 | 115.06 (26.43) | 69–130 | 6674 | 115.06 (26.43) | 25–180 | 1747 | 116.46 (12.77) | 92–135 | 6972 | 107.26 (27.34) | 5–172 | 22,809 | 109.13 (25.55) | 5–180 |
| Amphetamines | 6095 | 96.74 (22.85) | 12–144 | 4789 | 82.56 (19.31) | 54–111 | 4789 | 82.56 (19.31) | 18–123 | 1234 | 82.26 (13.35) | 66–110 | 4623 | 71.12 (17.69) | 6–101 | 17,431 | 83.40 (21.90) | 6–144 |
| Cannabinoids | 4528 | 71.87 (23.55) | 7–138 | 4172 | 71.93 (18.30) | 45–75 | 4172 | 71.93 (18.30) | 8–105 | 986 | 65.73 (10,62) | 46–85 | 3454 | 53.13 (16.46) | 1–87 | 13,652 | 65.32 (20.55) | 1–138 |
| Opioids | 1526 | 24.22 (6.86) | 5–42 | 1196 | 20.62 (7.44) | 10–34 | 1196 | 20.62 (7.44) | 1–40 | 257 | 17.13 (3.58) | 11–22 | 1077 | 16.56 (5.78) | 1–28 | 4226 | 20.22 (7.20) | 1–42 |
| Others a | 1446 | 22.95 (9.47) | 3–52 | 2213 | 38.1 5 (18.92) | 14–39 | 2213 | 38.15 (18.92) | 7–116 | 534 | 35.60 (9.26) | 23–57 | 1840 | 28.30 (10.92) | 0–66 | 6245 | 29.88 (14.38) | 0–116 |
| Location | ||||||||||||||||||
| Metropolitan | 11,396 | 180.88 (44.88) | 16–257 | 1365 | 170.62 (40.20) | 99–223 | 10,663 | 183.84 (40.18) | 33–284 | 2418 | 161.20 (16.04) | 136–193 | 9235 | 142.07 (34.94) | 7–196 | 35,077 | 167.83 (42.71) | 7–284 |
| Non‐metropolitan | 8732 | 138.60 (35.46) | 21–217 | 1110 | 138.75 (25.30) | 96–173 | 8221 | 141.74 (33.75) | 27–213 | 2295 | 152.00 (21.14) | 111–195 | 8646 | 133.01 (34.07) | 6–203 | 29,004 | 138.77 (33.51) | 6–217 |
| Episode completion (planned) | ||||||||||||||||||
| Overall | 8897 | 141.22 (51.29) | 27–249 | 1191 | 148.87 (32.77) | 90–197 | 10,998 | 189.62 (44.96) | 73–299 | 2554 | 170.26 (27.08) | 116–207 | 9286 | 142.86 (41.12) | 9–263 | 32,926 | 157.54 (48.92) | 9–299 |
| Gender | ||||||||||||||||||
| Male | 5768 | 91.55 (33.74) | 20–175 | 784 | 98.00 (22.32) | 58–119 | 7084 | 122.13 (27.71) | 53–182 | 1546 | 103.06 (17.39) | 81–136 | 5559 | 85.52 (26.22) | 3–159 | 20,741 | 99.23 (31.98) | 3–182 |
| Female | 3103 | 49.25 (18.75) | 6–82 | 405 | 50.62 (13.60) | 32–78 | 3881 | 66.91 (19.82) | 20–122 | 993 | 66.20 (15.97) | 34–94 | 3531 | 54.32 (17.48) | 5–103 | 11,913 | 57.00 (19.64) | 5–122 |
| Age group (years) | ||||||||||||||||||
| < 25 | 3406 | 54.06 (31.87) | 4–139 | 509 | 63.62 (18.01) | 32–86 | 4405 | 75.94 (28.63) | 32–177 | 879 | 58.60 (14.04) | 38–83 | 2747 | 42.26 (20.59) | 4–106 | 11,946 | 57.15 (29.23) | 4–177 |
| 25–59 | 5105 | 81.03 (23.89) | 10–137 | 609 | 76.12 (23.23) | 51–110 | 6095 | 105.08 (24.43) | 38–181 | 1582 | 105.46 (17.66) | 75–137 | 6307 | 97.03 (26.16) | 12–162 | 19,698 | 94.24 (26.33) | 10–181 |
| ≥ 60 | 320 | 5.07 (2.63) | 0–12 | 65 | 8.12 (3.87) | 4–16 | 432 | 7.44 (3.03) | 3–15 | 92 | 6.13 (2.42) | 1–9 | 480 | 7.38 (3.11) | 0–14 | 1389 | 6.64 (3.11) | 0–16 |
| Principal drug of concern | ||||||||||||||||||
| Alcohol | 3000 | 47.619 (17.11) | 7–82 | 421 | 52.62 (15.37) | 33–74 | 3990 | 68.80 (16.80) | 30–119 | 967 | 64.46 (11.81) | 49–87 | 3837 | 59.03 (17.07) | 5–99 | 12,215 | 58.44 (18.46) | 5–119 |
| Amphetamines | 2422 | 38.44 (13.09) | 9–70 | 291 | 36.37 (13.39) | 19–55 | 2451 | 42.25 (10.90) | 15–69 | 548 | 36.53 (9.19) | 21–54 | 2097 | 32.26 (10.22) | 3–56 | 7809 | 37.36 (11.95) | 3–70 |
| Cannabinoids | 1951 | 30.96 (17.85) | 2–81 | 300 | 37.50 (10.88) | 19–53 | 2340 | 40.34 (10.71) | 11–73 | 539 | 35.93 (8.92) | 22–52 | 1739 | 26.75 (9.84) | 0–63 | 6869 | 32.86 (13.98) | 0–81 |
| Opioids | 627 | 9.95 (3.94) | 1–24 | 66 | 8.25 (2.49) | 6–12 | 559 | 9.63 (4.36) | 2–19 | 105 | 7.00 (3.23) | 3–13 | 475 | 7.30 (3.71) | 0–19 | 1832 | 8.76 (4.06) | 0–24 |
| Others a | 733 | 11.63 (7.37) | 0–36 | 101 | 12.62 (5.40) | 4–21 | 1568 | 27.03 (16.69) | 6–92 | 359 | 23.93 (9.05) | 9–41 | 1097 | 16.87 (9.38) | 0–50 | 3858 | 18.45 (12.84) | 0–92 |
| Location | ||||||||||||||||||
| Metropolitan | 5090 | 80.79 (31.98) | 14–150 | 702 | 87.75 (24.09) | 46–114 | 6259 | 107.91 (28.64) | 46–202 | 1296 | 86.40 (19.08) | 53–134 | 4648 | 74.55 (24.05) | 5–151 | 18,193 | 87.04 (30.56) | 5–202 |
| Non‐metropolitan | 3655 | 58.01 (20.90) | 12–97 | 471 | 58.87 (14.61) | 41–78 | 4540 | 78.27 (20.37) | 26–121 | 1224 | 81.60 (18.11) | 56–120 | 4289 | 65.98 (20.76) | 4–125 | 14,179 | 67.84 (21.98) | 4–125 |
| Episode completion (unplanned) | ||||||||||||||||||
| Overall | 3654 | 58.00 (26.31) | 0–113 | 702 | 87.75 (16.96) | 54–108 | 4680 | 80.68 (23.33) | 27–142 | 1405 | 93.66 (27.45) | 72–182 | 5178 | 79.66 (28.65) | 3–164 | 15,619 | 74.73 (28.34) | 0–182 |
| Gender | ||||||||||||||||||
| Male | 2328 | 36.95 (17.60) | 0–74 | 468 | 58.50 (11.19) | 39–71 | 2974 | 51.27 (15.50) | 21–97 | 929 | 61.93 (22.72) | 41–136 | 3306 | 50.86 (19.73) | 1–109 | 10,005 | 47.87 (19.44) | 0–136 |
| Female | 1317 | 20.90 (9.68) | 0–45 | 234 | 29.25 (7.26) | 15–38 | 1685 | 29.05 (8.78) | 6–47 | 472 | 31.46 (6.89) | 23–46 | 1826 | 28.09 (10.90) | 0–59 | 5534 | 26.47 (10.24) | 0–59 |
| Age group (years) | ||||||||||||||||||
| < 25 | 1230 | 19.52 (10.10) | 0–50 | 233 | 29.125 (9.78) | 15–41 | 1468 | 25.31 (7.87) | 6–49 | 446 | 29.73 (6.95) | 18–44 | 1579 | 24.29 (11.93) | 1–53 | 4956 | 23.71 (10.37) | 0–53 |
| 25–59 | 2367 | 37.57 (18.16) | 0–75 | 454 | 56.75 (11.91) | 32–71 | 3099 | 53.43 (17.41) | 21–100 | 935 | 62.33 (23.09) | 45–139 | 3610 | 55.53 (21.32) | 2–119 | 10,465 | 50.071 (20.84) | 0–139 |
| ≥ 60 | 51 | 0.80 (0.99) | 0–4 | 15 | 1.875 (1.64) | 0–5 | 113 | 1.94 (1.97) | 0–9 | 33 | 2.20 (2.60) | 0–10 | 129 | 1.98 (1.77) | 0–9 | 341 | 1.63 (1.78) | 0–10 |
| Principal Drug of Concern | ||||||||||||||||||
| Alcohol | 1228 | 19.49 (10.19) | 0–47 | 228 | 28.5 (7.67) | 18–40 | 1363 | 23.50 (9.50) | 3–50 | 396 | 26.40 (5.88) | 19–37 | 1555 | 23.92 (9.09) | 0–51 | 4770 | 22.82 (9.56) | 0–51 |
| Amphetamines | 804 | 12.76 (6.91) | 0–31 | 177 | 22.125 (5.22) | 15–32 | 1128 | 19.44 (6.50) | 4–32 | 326 | 21.73 (5.92) | 12–31 | 1163 | 17.89 (7.71) | 0–36 | 3598 | 17.21 (7.57) | 0–36 |
| Cannabinoids | 264 | 4.19 (2.67) | 0–13 | 48 | 6.00 (5.24) | 0–17 | 400 | 6.89 (3.48) | 1–17 | 87 | 5.8 (3.14) | 2–13 | 322 | 4.95 (3.160) | 0–16 | 1121 | 5.36 (3.35) | 0–17 |
| Opioids | 996 | 15.80 (7.48) | 0–31 | 205 | 25.62 (10.89) | 13–40 | 1419 | 24.46 (8.18) | 8–46 | 449 | 29.93 (12.84) | 18–72 | 1667 | 25.64 (11.67) | 0–62 | 4736 | 22.66 (10.68) | 0–72 |
| Others a | 191 | 3.03 (2.08) | 0–8 | 37 | 4.62 (1.41) | 3–7 | 5.46 (2.58) | 1–13 | 116 | 7.73 (2.94) | 3–13 | 451 | 6.93 (3.67) | 0–22 | 1112 | 5.32 (3.28) | 0–22 | |
| Location | ||||||||||||||||||
| Metropolitan | 2088 | 33.14 (16.60) | 0–75 | 353 | 44.12 (14.23) | 27–68 | 2613 | 45.05 (14.11) | 15–88 | 715 | 47.66 (9.15) | 32–69 | 2608 | 40.12 (16.45) | 0–91 | 8377 | 40.08 (16.08) | 0–91 |
| Non‐metropolitan | 1461 | 23.19 (11.48) | 0–49 | 336 | 42.00 (15.07) | 21–68 | 1971 | 33.98 (14.94) | 7–103 | 668 | 44.53 (27.78) | 27–142 | 2509 | 38.60 (16.65) | 0–96 | 6945 | 33.22 (17.24) | 0–142 |
| Model | Noise parameters | Coefficient | Pre‐lockdowns (Intercept) | Lockdown 1 | Between lockdowns | Lockdown 2 | Post‐lockdown | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| β | SE | p | β | SE | p | β | SE | p | β | SE | p | β | SE | p | |||
| Commencements | |||||||||||||||||
| Total | |||||||||||||||||
| ARIMA(2,0,0)(1,0,0) | Intercept | 292.93 | 12.86 | < 0.001* | −814.85 | 490.04 | 0.098 | −14.52 | 39.30 | 0.712 | 317.30 | 398.25 | 0.427 | 54.45 | 56.86 | 0.339 | |
| Slope | 1.39 | 0.33 | < 0.001* | 11.14 | 7.26 | 0.127 | −0.79 | 0.49 | 0.105 | −3.47 | 2.92 | 0.237 | −1.71 | 0.45 | < 0.001* | ||
| Gender | |||||||||||||||||
| Male | ARIMA(1,0,0)(0,0,1) | Intercept | 192.11 | 8.79 | < 0.001* | −580.67 | 331.23 | 0.081 | 38.30 | 28.29 | 0.177 | 386.38 | 275.14 | 0.162 | 42.64 | 40.46 | 0.293 |
| Slope | 0.81 | 0.23 | < 0.001* | 8.00 | 4.91 | 0.105 | −0.96 | 0.35 | 0.006* | −3.56 | 2.02 | 0.08 | −1.13 | 0.32 | < 0.001* | ||
| Female | ARIMA(2,0,0)(1,0,0) | Intercept | 103.82 | 7.50 | < 0.001* | −217.30 | 232.56 | 0.351 | −57.10 | 23.34 | 0.015* | 16.49 | 208.52 | 0.937 | 41.18 | 33.88 | 0.226 |
| Slope | 0.50 | 0.20 | 0.011* | 2.84 | 3.46 | 0.413 | 0.25 | 0.29 | 0.382 | −0.51 | 1.53 | 0.741 | −0.71 | 0.27 | 0.009* | ||
| Age group | |||||||||||||||||
| < 25 years | ARIMA(1,0,0) | Intercept | 86.59 | 8.14 | < 0.001* | −104.70 | 279.22 | 0.708 | −50.68 | 26.54 | 0.058 | 320.92 | 238.06 | 0.179 | 144.15 | 38.26 | < 0.001* |
| Slope | 0.91 | 0.22 | < 0.001* | 0.81 | 4.15 | 0.846 | −0.08 | 0.33 | 0.821 | −3.16 | 1.75 | 0.072 | −1.76 | 0.30 | < 0.001* | ||
| 25–59 years | ARIMA(2,0,0)(1,0,0) | Intercept | 131.98 | 8.38 | < 0.001* | −612.12 | 337.36 | 0.071 | 77.03 | 27.26 | 0.005* | 170.79 | 272.43 | 0.531 | −32.27 | 39.02 | 0.409 |
| Slope | 1.76 | 0.27 | < 0.001* | 8.25 | 5.00 | 0.1 | −1.81 | 0.36 | < 0.001* | −2.43 | 2.01 | 0.227 | −1.19 | 0.35 | < 0.001* | ||
| ≥ 60 years | ARIMA(1,0,0) | Intercept | 9.34 | 1.25 | < 0.001* | −32.50 | 47.42 | 0.494 | −2.84 | 4.09 | 0.488 | 31.23 | 38.68 | 0.42 | −6.94 | 5.89 | 0.24 |
| Slope | 0.04 | 0.03 | 0.296 | 0.49 | 0.70 | 0.482 | 0.02 | 0.05 | 0.723 | −0.26 | 0.28 | 0.358 | 0.02 | 0.05 | 0.654 | ||
| Principal drug of concern | |||||||||||||||||
| Alcohol | ARIMA(1,0,0)(0,0,1) | Intercept | 96.62 | 5.23 | < 0.001* | −173.93 | 206.02 | 0.399 | −10.84 | 16.61 | 0.515 | 60.61 | 165.99 | 0.715 | 10.11 | 24.12 | 0.676 |
| Slope | 0.31 | 0.14 | 0.028* | 2.52 | 3.05 | 0.41 | 0.01 | 0.20 | 0.943 | −0.60 | 1.22 | 0.625 | −0.28 | 0.19 | 0.146 | ||
| Amphetamines | ARIMA(3,0,0)(1,0,0) | Intercept | 88.30 | 4.20 | < 0.001* | −338.44 | 160.21 | 0.036* | 20.34 | 13.52 | 0.134 | 124.09 | 131.19 | 0.345 | −19.06 | 19.45 | 0.328 |
| Slope | 0.32 | 0.11 | 0.005* | 4.68 | 2.38 | 0.05 | −0.56 | 0.17 | < 0.001* | −1.27 | 0.96 | 0.19 | −0.29 | 0.15 | 0.062 | ||
| Cannabinoids | ARIMA(5,0,0)(1,0,0) | Intercept | 52.99 | 4.32 | < 0.001* | 62.80 | 169.56 | 0.711 | 9.01 | 13.89 | 0.517 | 169.38 | 140.45 | 0.229 | 55.39 | 19.86 | 0.006* |
| Slope | 0.64 | 0.11 | < 0.001* | −1.37 | 2.51 | 0.588 | −0.53 | 0.17 | 0.002* | −1.77 | 1.03 | 0.087 | −0.93 | 0.16 | < 0.001* | ||
| Opioids | ARIMA(2,0,0)(1,0,0) | Intercept | 24.69 | 1.62 | < 0.001* | −77.35 | 60.87 | 0.205 | 11.56 | 5.24 | 0.028* | 9.07 | 50.12 | 0.857 | −13.37 | 7.53 | 0.077 |
| Slope | 0.01 | 0.04 | 0.858 | 1.10 | 0.90 | 0.225 | −0.16 | 0.07 | 0.015* | −0.13 | 0.37 | 0.725 | 0.03 | 0.06 | 0.65 | ||
| Other drugs | ARIMA(1,0,0)(0,0,1) | Intercept | 17.92 | 2.45 | < 0.001* | −121.19 | 102.95 | 0.24 | −43.13 | 7.95 | < 0.001* | 32.06 | 82.71 | 0.699 | 37.31 | 11.28 | 0.001* |
| Slope | 0.19 | 0.07 | 0.004* | 1.73 | 1.52 | 0.258 | 0.45 | 0.10 | < 0.001* | −0.29 | 0.61 | 0.628 | −0.34 | 0.09 | < 0.001* | ||
| Residential location | |||||||||||||||||
| Metropolitan | ARIMA(1,0,0) | Intercept | 161.38 | 9.02 | < 0.001* | −725.12 | 318.97 | 0.024* | −5.52 | 29.47 | 0.851 | 194.90 | 273.95 | 0.478 | −5.59 | 42.43 | 0.895 |
| Slope | 0.80 | 0.24 | 0.001* | 10.07 | 4.73 | 0.035* | −0.48 | 0.37 | 0.189 | −2.21 | 2.01 | 0.273 | −0.83 | 0.34 | 0.014* | ||
| Non‐metropolitan | ARIMA(1,0,0)(0,0,1) | Intercept | 121.75 | 7.13 | < 0.001* | −16.11 | 270.77 | 0.953 | −20.82 | 22.63 | 0.359 | 274.12 | 225.06 | 0.225 | 78.49 | 32.73 | 0.017* |
| Slope | 0.64 | 0.19 | < 0.001* | −0.09 | 4.01 | 0.982 | −0.20 | 0.28 | 0.481 | −2.42 | 1.65 | 0.145 | −0.99 | 0.26 | < 0.001* | ||
| Model | Noise parameters | Coefficient | Pre‐lockdowns (Intercept) | Lockdown 1 | Between lockdowns | Lockdown 2 | Post‐lockdown | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| β | SE | p | β | SE | p | β | SE | p | β | SE | p | β | SE | p | |||
| Planned cessations | |||||||||||||||||
| Total | |||||||||||||||||
| ARIMA(1,0,0)(0,0,1) | Intercept | 85.22 | 7.75 | < 0.001* | −142.10 | 322.70 | 0.66 | 17.53 | 24.65 | 0.478 | −199.07 | 252.68 | 0.432 | 182.33 | 35.43 | < 0.001* | |
| Slope | 1.91 | 0.20 | < 0.001* | 1.16 | 4.78 | 0.808 | −1.02 | 0.30 | < 0.001* | 0.18 | 1.86 | 0.923 | −2.59 | 0.28 | < 0.001* | ||
| Gender | |||||||||||||||||
| Male | ARIMA(3,0,0)(1,0,0) | Intercept | 56.63 | 4.68 | < 0.001* | −172.50 | 204.35 | 0.4 | 34.34 | 15.06 | 0.024* | −59.24 | 156.41 | 0.705 | 111.20 | 21.56 | < 0.001* |
| Slope | 1.20 | 0.12 | < 0.001* | 1.96 | 3.03 | 0.517 | −0.87 | 0.19 | < 0.001* | −0.42 | 1.15 | 0.712 | −1.64 | 0.17 | < 0.001* | ||
| Female | ARIMA(1,0,0)(0,0,1) | Intercept | 28.07 | 3.36 | < 0.001* | 21.93 | 139.63 | 0.875 | −16.84 | 10.60 | 0.114 | −157.30 | 109.84 | 0.154 | 91.08 | 15.25 | < 0.001* |
| Slope | 0.72 | 0.09 | < 0.001* | −0.69 | 2.07 | 0.739 | −0.15 | 0.13 | 0.243 | 0.71 | 0.81 | 0.377 | −1.07 | 0.12 | < 0.001* | ||
| Age group | |||||||||||||||||
| < 25 years | ARIMA(2,0,1)(1,0,0) | Intercept | 15.35 | 5.07 | 0.003* | −103.30 | 210.58 | 0.624 | −7.84 | 16.32 | 0.631 | −70.14 | 165.11 | 0.671 | 121.81 | 23.40 | < 0.001* |
| Slope | 1.26 | 0.14 | < 0.001* | 0.98 | 3.12 | 0.753 | −0.57 | 0.20 | 0.005* | −0.43 | 1.21 | 0.721 | −1.79 | 0.19 | < 0.001* | ||
| 25–59 years | ARIMA(1,0,0) | Intercept | 65.07 | 5.07 | < 0.001* | −37.62 | 205.24 | 0.855 | 36.68 | 16.67 | 0.029* | −134.78 | 162.10 | 0.407 | 31.63 | 23.88 | 0.187 |
| Slope | 0.59 | 0.14 | < 0.001* | 0.13 | 3.04 | 0.966 | −0.54 | 0.21 | 0.010* | 0.68 | 1.19 | 0.566 | −0.57 | 0.19 | 0.003* | ||
| ≥ 60 years | ARIMA(2,0,0) | Intercept | 3.59 | 0.58 | < 0.001* | 44.78 | 26.07 | 0.087 | −0.64 | 1.93 | 0.74 | 12.66 | 19.53 | 0.518 | 2.64 | 2.76 | 0.341 |
| Slope | 0.05 | 0.02 | 0.001* | −0.65 | 0.39 | 0.093 | −0.01 | 0.02 | 0.813 | −0.13 | 0.14 | 0.379 | −0.04 | 0.02 | 0.047* | ||
| Principal drug of concern | |||||||||||||||||
| Alcohol | ARIMA(1,0,0)(0,0,1) | Intercept | 29.78 | 3.12 | < 0.001* | −83.49 | 129.60 | 0.52 | 14.36 | 10.07 | 0.155 | −147.60 | 101.69 | 0.148 | 80.39 | 14.44 | < 0.001* |
| Slope | 0.61 | 0.08 | < 0.001* | 0.97 | 1.92 | 0.614 | −0.35 | 0.13 | 0.005* | 0.72 | 0.75 | 0.334 | −0.89 | 0.11 | < 0.001* | ||
| Amphetamines | ARIMA(1,0,1)(1,0,1) | Intercept | 28.05 | 2.35 | < 0.001* | 114.69 | 97.52 | 0.241 | 25.07 | 7.88 | 0.002* | 56.07 | 75.80 | 0.46 | 9.84 | 11.24 | 0.382 |
| Slope | 0.36 | 0.06 | < 0.001* | −1.94 | 1.45 | 0.181 | −0.46 | 0.10 | < 0.001* | −0.71 | 0.56 | 0.204 | −0.39 | 0.09 | < 0.001* | ||
| Cannabinoids | ARIMA(2,0,1)(1,0,0) | Intercept | 10.49 | 2.24 | < 0.001* | −21.97 | 95.00 | 0.817 | 11.44 | 7.30 | 0.119 | −131.59 | 73.37 | 0.074 | 48.45 | 10.47 | < 0.001* |
| Slope | 0.69 | 0.06 | < 0.001* | 0.04 | 1.41 | 0.976 | −0.50 | 0.09 | < 0.001* | 0.46 | 0.54 | 0.396 | −0.86 | 0.08 | < 0.001* | ||
| Opioids | ARIMA(2,0,0)(1,0,0) | Intercept | 10.14 | 1.07 | < 0.001* | −23.32 | 39.79 | 0.558 | 2.33 | 3.42 | 0.497 | −9.10 | 32.88 | 0.782 | 3.60 | 4.95 | 0.467 |
| Slope | 0.00 | 0.03 | 0.926 | 0.32 | 0.59 | 0.591 | −0.03 | 0.04 | 0.497 | 0.04 | 0.24 | 0.867 | −0.04 | 0.04 | 0.347 | ||
| Other drugs | ARIMA(1,0,0)(0,0,1) | Intercept | 4.07 | 2.22 | 0.068 | −89.60 | 93.00 | 0.336 | −27.35 | 7.00 | < 0.001* | −23.70 | 74.65 | 0.751 | 42.30 | 10.07 | < 0.001* |
| Slope | 0.26 | 0.06 | < 0.001* | 1.19 | 1.38 | 0.39 | 0.24 | 0.09 | 0.006* | 0.06 | 0.55 | 0.91 | −0.42 | 0.08 | < 0.001* | ||
| Residential location | |||||||||||||||||
| Metropolitan | ARIMA(1,0,0)(0,0,1) | Intercept | 47.04 | 5.42 | < 0.001* | −245.03 | 223.37 | 0.274 | 6.50 | 17.37 | 0.709 | −219.40 | 175.88 | 0.214 | 84.75 | 24.91 | < 0.001* |
| Slope | 1.15 | 0.14 | < 0.001* | 3.08 | 3.31 | 0.352 | −0.59 | 0.21 | 0.007* | 0.75 | 1.29 | 0.563 | −1.45 | 0.20 | < 0.001* | ||
| Non‐metropolitan | ARIMA(1,0,0) | Intercept | 35.72 | 3.70 | < 0.001* | 113.73 | 154.49 | 0.462 | 16.00 | 12.18 | 0.19 | 22.08 | 119.99 | 0.854 | 99.91 | 17.46 | < 0.001* |
| Slope | 0.76 | 0.10 | < 0.001* | −2.11 | 2.29 | 0.357 | −0.48 | 0.15 | 0.002* | −0.59 | 0.88 | 0.505 | −1.14 | 0.14 | < 0.001* | ||
| Model | Noise parameters | Coefficient | Pre‐lockdowns (Intercept) | Lockdown 1 | Between lockdowns | Lockdown 2 | Post‐lockdown | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| β | SE | p | β | SE | p | β | SE | p | β | SE | p | β | SE | p | |||
| Unplanned cessations | |||||||||||||||||
| Total | |||||||||||||||||
| ARIMA(1,0,0) | Intercept | 34.63 | 6.85 | < 0.001* | −92.40 | 254.66 | 0.717 | 63.34 | 22.41 | 0.005* | −121.19 | 209.39 | 0.563 | 135.96 | 32.27 | < 0.001* | |
| Slope | 0.79 | 0.19 | < 0.001* | 1.40 | 3.78 | 0.711 | −0.95 | 0.28 | < 0.001* | 0.53 | 1.54 | 0.732 | −1.29 | 0.26 | < 0.001* | ||
| Gender | |||||||||||||||||
| Male | ARIMA(1,0,0) | Intercept | 20.41 | 4.49 | < 0.001* | −122.04 | 171.91 | 0.479 | 49.55 | 14.72 | < 0.001* | −161.00 | 139.41 | 0.249 | 102.46 | 21.18 | < 0.001* |
| Slope | 0.55 | 0.12 | < 0.001* | 1.84 | 2.55 | 0.472 | −0.73 | 0.18 | < 0.001* | 0.92 | 1.02 | 0.367 | −0.95 | 0.17 | < 0.001* | ||
| Female | ARIMA(2,0,0)(0,0,1) | Intercept | 14.33 | 2.48 | < 0.001* | 2.97 | 95.40 | 0.975 | 11.16 | 8.00 | 0.165 | 64.02 | 78.77 | 0.417 | 30.84 | 11.52 | 0.008* |
| Slope | 0.23 | 0.07 | < 0.001* | −0.05 | 1.41 | 0.974 | −0.19 | 0.10 | 0.053 | −0.57 | 0.58 | 0.326 | −0.32 | 0.09 | < 0.001* | ||
| Age group | |||||||||||||||||
| < 25 years | ARIMA(2,0,0)(1,0,1) | Intercept | 10.20 | 2.86 | < 0.001* | −186.31 | 100.12 | 0.064 | 23.45 | 8.83 | 0.008* | 20.94 | 84.89 | 0.805 | 64.24 | 12.88 | < 0.001* |
| Slope | 0.27 | 0.07 | < 0.001* | 2.76 | 1.49 | 0.064 | −0.35 | 0.11 | 0.001* | −0.28 | 0.62 | 0.657 | −0.55 | 0.10 | < 0.001* | ||
| 25–59 years | ARIMA(1,0,0)(0,0,1) | Intercept | 22.48 | 5.25 | < 0.001* | 57.61 | 196.63 | 0.77 | 33.54 | 17.03 | 0.05 | −65.14 | 166.81 | 0.697 | 49.14 | 24.27 | 0.044* |
| Slope | 0.51 | 0.14 | < 0.001* | −0.81 | 2.92 | 0.783 | −0.53 | 0.21 | 0.013* | 0.27 | 1.22 | 0.828 | −0.59 | 0.19 | 0.002* | ||
| ≥ 60 years | ARIMA(1,0,0) | Intercept | 0.28 | 0.40 | 0.493 | 3.74 | 16.73 | 0.824 | −0.22 | 1.33 | 0.868 | 8.47 | 13.07 | 0.518 | −0.46 | 1.91 | 0.809 |
| Slope | 0.02 | 0.01 | 0.093 | −0.05 | 0.25 | 0.84 | 0.00 | 0.02 | 0.964 | −0.07 | 0.10 | 0.491 | −0.01 | 0.02 | 0.702 | ||
| Principal drug of concern | |||||||||||||||||
| Alcohol | ARIMA(2,0,0)(1,0,0) | Intercept | 25.56 | 7.00 | < 0.001* | −163.13 | 121.33 | 0.18 | 44.02 | 22.22 | 0.049* | 13.38 | 136.96 | 0.922 | 44.21 | 35.83 | 0.219 |
| Slope | −0.06 | 0.20 | 0.759 | 2.82 | 1.82 | 0.122 | −0.28 | 0.31 | 0.36 | 0.00 | 1.01 | 0.998 | −0.16 | 0.29 | 0.569 | ||
| Amphetamines | ARIMA(2,0,0) | Intercept | 5.14 | 1.50 | < 0.001* | −75.23 | 62.22 | 0.228 | 25.41 | 4.93 | < 0.001* | 18.36 | 48.84 | 0.707 | 47.04 | 7.08 | < 0.001* |
| Slope | 0.25 | 0.04 | < 0.001* | 1.11 | 0.92 | 0.228 | −0.36 | 0.06 | < 0.001* | −0.27 | 0.36 | 0.46 | −0.44 | 0.06 | < 0.001* | ||
| Cannabinoids | ARIMA(2,0,0) | Intercept | 2.77 | 0.86 | 0.001* | −42.45 | 32.29 | 0.19 | 5.05 | 2.82 | 0.075 | 11.99 | 26.48 | 0.651 | 9.61 | 4.05 | 0.019* |
| Slope | 0.05 | 0.02 | 0.038* | 0.63 | 0.48 | 0.188 | −0.06 | 0.04 | 0.106 | −0.11 | 0.19 | 0.56 | −0.09 | 0.03 | 0.006* | ||
| Opioids | ARIMA(1,0,0)(0,0,1) | Intercept | 10.27 | 2.50 | < 0.001* | 52.97 | 95.86 | 0.581 | 14.98 | 7.77 | 0.055 | 5.82 | 82.40 | 0.944 | 36.29 | 11.20 | 0.001* |
| Slope | 0.19 | 0.06 | 0.003* | −0.75 | 1.42 | 0.6 | −0.20 | 0.10 | 0.039* | −0.09 | 0.60 | 0.884 | −0.31 | 0.09 | < 0.001* | ||
| Other drugs | ARIMA(1,0,0) | Intercept | 1.85 | 0.71 | 0.010* | −13.33 | 28.66 | 0.642 | 1.78 | 2.35 | 0.449 | −25.12 | 22.74 | 0.27 | 11.06 | 3.37 | 0.001* |
| Slope | 0.04 | 0.02 | 0.034* | 0.20 | 0.42 | 0.642 | −0.02 | 0.03 | 0.451 | 0.18 | 0.17 | 0.269 | −0.07 | 0.03 | 0.006* | ||
| Residential location | |||||||||||||||||
| Metropolitan | ARIMA(1,0,1)(1,0,1) | Intercept | 17.45 | 3.74 | < 0.001* | 7.86 | 143.15 | 0.956 | 40.44 | 12.52 | 0.001* | −29.08 | 115.96 | 0.802 | 92.76 | 17.86 | < 0.001* |
| Slope | 0.52 | 0.10 | < 0.001* | −0.23 | 2.12 | 0.915 | −0.64 | 0.16 | < 0.001* | −0.09 | 0.85 | 0.915 | −0.91 | 0.14 | < 0.001* | ||
| Non‐metropolitan | ARIMA(1,0,0)(0,0,1) | Intercept | 14.75 | 3.88 | < 0.001* | −157.76 | 158.56 | 0.321 | 23.37 | 12.55 | 0.064 | −83.70 | 126.72 | 0.51 | 40.92 | 18.01 | 0.024* |
| Slope | 0.30 | 0.10 | 0.005* | 2.44 | 2.35 | 0.3 | −0.33 | 0.16 | 0.035* | 0.53 | 0.93 | 0.566 | −0.39 | 0.14 | 0.007* | ||
3.1Overall
3.1.1Commencements
A total of 20,800 episodes of care commenced during the pre‐lockdowns period (Table 1). Pre‐lockdowns, there was a significant upward trend of 1.39 episode commencements/week. There were no significant level or trend changes at lockdown one, between lockdowns, or at lockdown two. Post‐lockdowns, there was a significant reversal in trend, with commencements declining at a rate of 1.71 episodes per week.
3.1.2Cessations
3.1.2.1Planned Cessations
A total of 8897 episodes of care were ceased as planned during the pre‐lockdowns period (Table 1). Planned cessations had a significant upward trend of 1.91 episodes/week pre‐lockdowns. Between lockdowns, planned cessations declined by 1.02 episodes/week. Post‐lockdowns, there was a significant increase in the level of planned cessations, with a declining trend thereafter at a rate of 2.59 episodes per week.
3.1.2.2Unplanned Cessations
A total of 3654 unplanned cessations occurred during the pre‐lockdowns period. Pre‐lockdowns, unplanned cessations had a significant upward trend of 0.79 episodes/week. Between lockdowns, there was a significant increase in the level of unplanned cessations, with a subsequent significant downward trend of 0.95 episodes per week. Post‐lockdowns there was a further significant increase in the level of unplanned cessations and a steeper downward trend of 1.29 episodes per week.
3.2Subgroup Analyses
3.2.1Commencements
Detailed subgroup analyses are presented in Appendix 3. Principal drug of concern and residential location were the only subgroup analyses to demonstrate a shift in the number of episodes that commenced during either lockdown period. At lockdown one, there was a significant decrease in the level of commencements for amphetamines, followed by an increase in commencements/week across lockdown one. Lockdown one also saw a significant reduction in the level of commencements for metropolitan residents and an increasing trend in episodes/week.
3.2.2Cessations
None of the subgroup analyses found evidence of a shift in the number of cessations (planned or unplanned) that occurred during either lockdown period. Detailed subgroup analyses are presented in Appendix 3.
4Discussion
The current study investigated the effect of the lockdown measures during COVID‐19 on service delivery among a state‐wide sample of outpatient non‐government AOD services in NSW, Australia, and whether there were differential impacts by gender, age, principal drug of concern and residential location. In the lead up to lockdown one we saw an increase in the number of episodes that commenced and ceased (planned and unplanned) and a corresponding decline once the lockdown measures were ultimately lifted. Contrary to expectation, although episode commencements and cessations (planned and unplanned) appeared to decline during the first week of each lockdown, the recovery was such that overall, neither of these periods appreciably differed from the pre‐lockdowns period. For the most part, this finding was the same irrespective of participant gender, age, principal drug of concern and residential location. Surprisingly, we found evidence that commencements increased during lockdown one for people with amphetamines as their principal drug of concern and for metropolitan participants. Although we cannot directly comment on whether episodes were delivered face‐to‐face or via telehealth, given the widespread restrictions around movement and gatherings, our findings point to the likely role of telehealth in sustaining (and for some groups, perhaps temporarily increasing) service provision during the COVID‐19 lockdowns.
Prior studies have similarly demonstrated an increase in service utilization early in the COVID‐19 pandemic (Palzes et al. 2022; Yang et al. 2020). This apparent increased demand for services is likely driven by elevated rates of distress and substance use (Roberts et al. 2021) and changes to the availability of alcohol and other drugs (Farhoudian et al. 2021). This finding is important when interpreting our results, since all our comparisons are made relative to this period. Accordingly, the absence of a significant difference between this period and the number of treatment episodes that commenced during and between lockdowns may suggest that services were able to continue meeting this increased demand despite the impact of COVID‐19. Aligned with this interpretation, Australia saw increases in both the number of mental health care services delivered in 2020, and the utilization of telehealth (Australian Institute of Health and Welfare 2021). However, contrary to our findings, a prior report identified a reduction in service utilization within the NSW non‐government AOD sector specifically during the first year of the COVID‐19 pandemic (Van De Ven et al. 2021). Given that the impact of COVID‐19 on service utilization has been found to vary according to service type (Lin et al. 2023), this discrepancy is likely due to differences in the population studied (i.e., new episodes across all AOD services (Van De Ven et al. 2021) vs. all episodes within community AOD services in the current study). Indeed, relative to community‐based services, this prior report demonstrated that residential rehabilitation services were “more severely constrained in their ability to provide treatment” (Van De Ven et al. 2021). It is also important to consider differences in study methods, with our findings derived from a 209‐week period that extends beyond the first year of the COVID‐19 pandemic and analysis methods that account for pre‐existing trends in service utilization.
Despite published evidence for the increased utilization of telehealth to deliver AOD treatment during the COVID‐19 pandemic (Vakkalanka et al. 2024), we did not see a corresponding increase in the overall number of community treatment episodes that commenced during the post‐lockdown period. Our findings would therefore suggest that telehealth was being used in lieu of face‐to‐face services. Findings from Australia (Yeatman et al. 2023; Carlyle et al. 2023) and abroad (Palzes et al. 2022) similarly demonstrate that increases in telehealth delivered services are proportional to a decline in face‐to‐face service provision. Although accumulating evidence from both clients (Sugarman et al. 2021) and clinicians (Mark et al. 2021) supports the feasibility of telehealth for delivering addiction services, attitudes regarding face‐to‐face service provision as the “gold‐standard” persist (Thomas et al. 2022). To optimize the potential of telehealth, education and training are therefore essential (Appleton et al. 2023) (Australian Digital Health Agency 2023). Careful workforce planning is also required to ensure that telehealth is culturally safe and used to augment rather than replace face‐to‐face service provision (Mathew et al. 2023). To ensure that health inequalities are not exacerbated, novel solutions (Appleton et al. 2023) to provide low cost access to reliable infrastructure would also be of benefit.
When data were examined separately according to gender, age, principal drug of concern and residential location, for the most part, neither lockdown period appreciably differed from the pre‐lockdown period with regard to either commencements or cessations (planned or unplanned). A criticism of transitioning from predominantly face‐to‐face service provision to telehealth is that it may disproportionately disadvantage vulnerable populations already at risk of poorer outcomes (Vakkalanka et al. 2024). Older people and those living in rural regions are among the vulnerable populations at greater risk for this so‐called “digital divide” (Vakkalanka et al. 2024). The apparent maintenance of service provision for people aged 60 and above and those in rural locations observed during both lockdown periods is therefore reassuring.
Although service provision did not appear to decline for participants in non‐metropolitan regions, it also did not increase above that of pre‐lockdowns. During the first lockdown, there was an increase in commencements/week for people in metropolitan (but not non‐metropolitan) locations. Although further evidence is needed, this finding may reflect regional differences in infrastructure. For example, the infrastructure required for seamless provision of video consultations (e.g., high‐speed internet connectivity) is more readily available in metropolitan locations. This likely affords metropolitan participants with greater choice over telehealth modality, with both choice (Savira et al. 2023) and video consultations (Savira et al. 2024) linked to greater utilization of telehealth within healthcare more broadly. However, efforts to determine whether these findings generalize to AOD settings are needed. Improved understanding of the potential of telehealth for improving accessibility of AOD treatment, particularly among those outside of metropolitan locations, is also needed.
There was also a trend for commencements/week to increase during lockdown one for people with amphetamines as their principal drug of concern (but not other substances). This may reflect increased demand for treatment due to COVID‐19‐related changes in the drug market. For example, Australia saw a profound reduction in the availability of methamphetamine, and a corresponding escalation in cost (Rathnayake et al. 2023). However, given that COVID‐19 also precipitated changes in the availability and cost of a range of substances (Rathnayake et al. 2023; Dietze and Peacock 2020), it is unclear why we only found evidence for a change in episode commencements for amphetamines. Perhaps the intensity of the withdrawal profile of methamphetamine (Cruickshank and Dyer 2009) in concert with limited availability of other substances increased the urgency of help‐seeking. There is also evidence to suggest that telehealth shows promise for overcoming the range of help‐seeking barriers encountered by people who use methamphetamine (Rubenis et al. 2021).
Rather than during the lockdowns, it was in between the lockdown periods that subgroup differences were most apparent. Firstly, the number of episodes for women remained steady, whereas men exhibited a significant decline. This may reflect gender differences in the utilization of telehealth services, with some evidence to suggest that women may be more likely to use this treatment modality (Vakkalanka et al. 2024). Secondly, the number of episodes for younger and older people remained steady, whereas a decline was observed among people aged 25–59. Age‐related heterogeneity in the impact of COVID‐19 on community‐based AOD treatment has been demonstrated previously (Palzes et al. 2022) and may reflect age‐related differences in substance use patterns, help‐seeking or engagement with telehealth. Thirdly, aligned with published accounts (Busch et al. 2023), the number of alcohol episodes remained steady. In contrast, the number of commencements/week for amphetamines, cannabinoids and opioids declined, whereas “other” drugs increased. Published evidence is similarly mixed, with the impact of COVID‐19 on service provision shown to differ according to the type (Carlyle et al. 2021) and number of substances used (Mellis et al. 2021) and type of service sought (Cooper et al. 2023; Saloner et al. 2022; Schofield et al. 2022). The return to face‐to‐face services during this period may also have contributed to the observed subgroup differences.
4.1Strengths and Limitations
Our findings are unique in that they are derived from a large sample of longitudinal data spanning almost 4 years. Data reflect the number of treatment episodes that commenced and ceased (planned and unplanned) before, during and after the COVID‐19 pandemic, allowing us to look at the long‐term effects of COVID‐19 on community AOD service utilization. As the data were routinely captured by a range of community‐based service providers, our sample is diverse and reflects routine service provision for a range of presenting concerns. Findings were analyzed using interrupted time‐series modeling to control for complex, non‐linear trends including seasonality, thereby more clearly demonstrating the unique contribution of COVID‐19 to changes in service utilization.
Our findings should also be considered in light of several limitations. Firstly, as our data are at the level of service episode, not individual client‐level, we are unable to differentiate between unique and repeat presentations. It is therefore unclear whether our findings reflect use of telehealth by people who had previously used face‐to‐face services, engagement of people attending services for the first time, or a combination thereof. Secondly, modality of service provision is not routinely captured as part of the National Minimum Data Set collected by AOD services in Australia. We are therefore unable to disaggregate data into face‐to‐face or telehealth to directly examine the impact of COVID‐19 on mode of service delivery. Thirdly, our findings are based on data captured by non‐government providers of community‐based treatment in NSW, Australia. Findings therefore may not be representative of the impact of COVID‐19 in other states (e.g., Victoria) that underwent more frequent, longer and stringent periods of restriction, nor of AOD treatment provided outside of the non‐government sector, different funding models or residential settings. It is also important to note that although routinely captured data provide important insights into service provision, it may also be limited due to clinician/service level errors or omissions in data entry. Our location variable was condensed from five levels of remoteness to a dichotomous variable representing metropolitan and non‐metropolitan locations. Although this was necessary to account for small numbers, particularly in remote and very remote regions, it may obscure findings within these regions. Data were not analyzed according to ethnicity and we are therefore unable to comment whether the current findings generalize to the experience of First Nations peoples. Finally, the analysis plan was not pre‐registered, and as such the results should be considered exploratory.
5Conclusions
Our findings indicate that the introduction of COVID‐19 lockdown measures did not adversely impact the number of treatment episodes delivered within a diverse sample of community‐based AOD services in NSW, Australia. These disruptions were not associated with increases in unplanned dropout among those accessing treatment, and no adverse impacts on access were found among key priority groups. Given widespread restrictions in movement and gathering precipitated by COVID‐19, this observed maintenance in service provision likely reflects the rapid transition to telehealth and highlights the resilience of the non‐government AOD sector for sustaining care under challenging circumstances.
Funding
The study was supported by funds from the NSW Ministry of Health via the Network of Alcohol and other Drugs Agencies. AKB and GC are funded by the National Health and Medical Research Council Meaningful Outcomes in Substance Use Treatment Centre of Research Excellence. LH is funded by the National Centre for Youth Substance Use Research (NCYSUR). NCYSUR is supported by funding from the Australian Government Department of Health, under the Drug and Alcohol Program.
Disclosure
M.L.L. and R.S. are employed by the Network of Alcohol and Other Drug Agencies (NADA).
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Acknowledgments
The authors acknowledge all staff and clinicians who collect and report data for the NADABase, as well as NADA as custodians of the data. The authors also acknowledge all clients of NADA member services whose data were used as part of this study. Open access publishing facilitated by University of Wollongong, as part of the Wiley ‐ University of Wollongong agreement via the Council of Australasian University Librarians.
No AI tools or technologies were used in the drafting of this manuscript. ChatGPT was used during the submission process to ask for advice on how the abstract could be condensed from 318 to 300 words without changing the meaning or tone. Selected suggestions were edited and incorporated. The authors reviewed the abstract prior to submission and take full responsibility for the content.
Data Availability Statement
Access to the data that support the findings of this study is subject to third party restrictions and may be requested from the Network of Alcohol and Other Drug Agencies.
Appendix Group
Operationalization of Study Periods
Study interval Study dates Study week Interval duration (weeks) Operationalization [9, 10, 12] Pre‐lockdowns January 1, 2019–March 16, 2020 1–63 63 — Lockdown 1 March 17, 2020–May 11, 2020 64–71 8 Restrictions around movement and gatherings introduced on March 18, 2020, and lifted on May 15, 2020. As restrictions commenced during week 64 and were lifted during Week 72 of the study, lockdown one is defined as Week 64–71 Between lockdowns May 12, 2020–June 21, 2021
a
72–129 58 — Lockdown 2 June 22, 2021–October 4, 2021 130–144 15 Restrictions around movement and gatherings introduced on June 25, 2021, and lifted on October 10, 2021. As restrictions commenced during Week 130 and were lifted during Week 145 of the study, lockdown two is defined as Week 130–144 Post‐lockdowns October 5, 2021–January 2, 2023 145–209 65 —
Assumption Checking
Autocorrelations in the residuals for both 104‐lags (twice the seasonality) and 10‐lags were tested using the Ljung‐Box and Box‐Pierce tests. Additional assumption checks for normality and homoscedasticity were also performed. Portmanteau tests indicated that the ARIMA model successfully addressed the autocorrelations in nearly all time series assessed. Residual autocorrelations (considering 10 lags) were still indicated for commencement counts of the 25–59‐year age group and unplanned alcohol cessations on both Ljung‐Box test and Box‐Pierce tests. Residual diagnostic plots indicated that residuals were reasonably approximated by a normal distribution on most time series, although a slight positive skew in residuals was indicated for time series with low estimated counts, particularly regarding unplanned cessations. Specifically, the Kolmogorov–Smirnov test indicated significant deviations from the standard normal distribution for unplanned total cessations (p = 0.022), unplanned rural cessations (p = 0.001), unplanned other drug cessations (p = 0.011), unplanned 25–59‐year‐old cessations (p = 0.029), and planned total cessations (p = 0.037). Considering the total number of time series modeled (39), the diagnostics indicate that the ARIMA‐based approach was mostly successful in addressing the autocorrelations and treatment of the residuals.
Detailed Subgroup Analysis
Commencements
Gender
Pre‐lockdowns, both genders showed a significant upward trend in commencements/week, followed by a decline for males between lockdowns and a decline post‐lockdown for both genders.
Age
Pre‐lockdowns, there was a significant upward trend in commencements/week for people younger than 25 and 25–59 years. Between lockdowns, there was a significant reduction in the level of commencements for 25–59‐year‐olds, with a significantly declining trend in commencements/week, which further declined post‐lockdowns. For the under 25 years, there was a significant increase in the level of commencements post‐lockdowns, and a decline in commencements/week.
Principal Drug of Concern
Pre‐lockdowns, there was a significant upward trend in commencements/week for all substances except opioids. At lockdown one there was a significant decrease in the level of commencements for amphetamines, followed by an increase in commencements/week across lockdown one. Between lockdowns, there was a significant decrease in the level of commencements, and a decline in commencements/week for opioids and other drugs. Amphetamines and cannabinoids also showed a significant decline in commencements/week between lockdowns. Post‐lockdown there was a significant increase in the level of commencements for cannabinoids and other drugs and a subsequent decline in the trend.
Residential Location
Pre‐lockdowns both metropolitan and non‐metropolitan residents had a significant upward trend in commencements/week. At lockdown one there was a significant reduction in the level of commencements for metropolitan residents and an increasing trend in episodes/week. Post‐lockdown there was a significant reduction in the level of commencements for non‐metropolitan residents and a declining trend in commencements/week for both groups.
Planned Cessations
Gender
For both genders, there was a significant upward trend in planned cessations/week across the pre‐lockdown period. Between lockdowns, males had a significant reduction in the level of planned cessations and a declining trend in planned cessations/week. Post‐lockdowns both genders showed a significant increase in the level of planned cessations and a declining trend in planned cessations/week.
Age
Pre‐lockdowns, all age groups showed a significant upward trend in planned cessations/week. Between lockdowns, there was a significant reduction in the level of planned cessations for 25–59‐year‐olds, and a decline in planned cessations/week between lockdowns for both 25–59 and < 25‐year‐olds. Post‐lockdowns there was a significant increase in the number of planned cessations for < 25‐year‐olds, and a declining trend in planned cessations/week for all age groups.
Principal Drug of Concern
All substance types, except opioids, showed a significant upward trend in planned cessations pre‐lockdowns. Between lockdowns, there was a significant reduction in the level of planned cessations for amphetamines and “other drugs.” There was also a declining trend in planned cessations/week for amphetamines, alcohol and cannabinoids but an increasing trend for “other drugs.” Post‐lockdowns there was a significant increase in the level of planned cessations for alcohol, cannabinoids and other drugs, and a declining trend for all groups except opioids.
Residential Location
For both metropolitan and non‐metropolitan areas, planned cessations had a significant upward trend pre‐lockdowns, followed by a significant reversal of this trend between lockdowns. Post‐lockdowns there was a significant decrease in the level of planned completions for both groups and a significant declining trend in planned cessations/week.
Unplanned Cessations
Gender
Pre‐lockdowns, there was a significant upward trend in unplanned cessations for both genders. Between lockdowns, males had a significant increase in the level of unplanned cessations, and a declining trend in unplanned cessations/week. Post‐lockdown there was a significant increase in the level of unplanned cessations for both genders, and a declining trend in unplanned cessations/week.
Age
Pre‐lockdowns there was a significant upward trend in unplanned cessations for age groups up to 59 years. Between lockdowns, there was an increase in the level of unplanned cessations for < 25‐year‐olds, and a declining trend in unplanned cessations/week. Post‐lockdowns, the age groups up to 59 years showed a significant increase in the level of unplanned cessations, and a declining rate in unplanned cessations/week.
Principal Drug of Concern
Unplanned cessations had a significant upward trend pre‐lockdowns for all substances except alcohol. Between lockdowns, there was a significant increase in the level of unplanned cessations for alcohol and amphetamines. There was also a declining trend in unplanned cessations/week for amphetamines and opioids. Post‐lockdown, all substances except alcohol had a significant increase in the level of unplanned cessations and a declining trend in unplanned cessations/week.
Residential Location
Pre‐lockdowns, both regions demonstrated a significant upward trend in unplanned cessations. Between lockdowns, there was a significant increase in the level of unplanned cessations for metropolitan residents. There was also a declining trend of unplanned cessations/week for both groups. Post‐lockdowns, there was a significant increase in the level of unplanned cessations for both groups, and declining trends in unplanned cessations/week.