The elective care waiting list: insights from linked data
Key points
- Reducing the NHS elective waiting list is a major national priority. But relatively little is known about the impact of long waiting lists on patient health or who is most at risk of adverse events while waiting. The current absence of a national, patient-level, linked dataset created an opportunity for the Networked Data Lab (NDL) to use local datasets to address evidence gaps.
- This briefing presents NDL analysis on the elective care waiting list. Led by the Health Foundation, the NDL is a network of analytical teams across the UK. NDL teams from four areas (Cheshire and Merseyside, Grampian, North West London and West Yorkshire) accessed, linked and analysed local data sources to explore how long different population groups wait for elective care; the reasons people leave the waiting list; and the health care use of those waiting.
- Wait times vary across population groups and between specialties. Patients from more deprived areas tend to wait longer than those from more affluent areas, and people of Asian, Black or mixed backgrounds were observed to have longer median wait times than white patients. Analysis performed in one area showed people waiting for procedures in multiple different specialties also waited for longer.
- Waits ended with receipt of a first treatment in 47% of cases, while 30% ended due to a decision not to treat. People from more deprived areas more frequently had pathways end due to missed appointments, likely indicating greater barriers to attendance. We could only analyse how waiting periods ended in one NDL location (Cheshire and Merseyside), due to data-quality issues in other locations.
- GP appointments and issuing prescriptions were by far the most common health care events among patients waiting. People on the elective waiting list were observed to have between 9 and 14 GP appointments per year. Despite much lower unit costs, the weekly cost of GP appointments for patients on the waiting list in NDL areas (£547 per week per 100 people waiting) far outstripped the weekly cost of A&E attendances (£251 per week per 100 people waiting).
- People from more deprived areas were observed to use more health care than people from less deprived areas. People from more deprived areas had higher rates of A&E attendances and emergency admissions than people from less deprived areas before, during and after waiting.
- Use of health care overall is typically higher in the immediate months before and after being on the waiting list than it is while waiting. Those waiting for longer also had relatively greater increases in GP appointments and A&E attendances following treatment, compared with those waiting for less time.
- Data relating to the waiting list often have substantial limitations and quality issues. This limits possibilities for linkage, hampering our ability to effectively build a picture of the waiting list or the effects of waiting on individuals and the health system. More accurate and consistent data and recording practices are imperative to manage the length of elective waits more effectively.
Figure 1
Continued improvement, however, will depend on how fast both referrals and treatments increase. Recent Health Foundation analysis suggests that although meeting the 18-week target by 2029 is possible, doing so will nonetheless be a challenge. Yet despite political consensus around tackling long waits, and its prominence in the government’s health mission, relatively little is known about the full impact of long NHS waiting lists on patient health, or who is most at risk of severe consequences. Variation in waiting time between population groups was also relatively underexplored until NHS England’s recent data release on the subject. This found that people from the most deprived areas and people of Asian or Asian British backgrounds were more likely to wait for longer than 18 weeks than any other group.
Very little analysis is available that takes a system-wide view of patients’ use of health care while on the waiting list or the costs of this care. This is in part because linked datasets are not widely accessible and the national-level Waiting List Minimum Dataset (WLMDS) is currently unavailable for research use. This created an opportunity for the Health Foundation’s NDL Labs to use their local datasets to generate insights addressing these evidence gaps.
About this briefing
This briefing presents analyses from the Health Foundation’s Networked Data Lab (NDL) on the impact of long waits for elective care. The NDL is a collaborative network of analytical teams from across the UK who create and use linked datasets to produce novel analytical insights into health and care services.
Here, we present analyses from four areas of the UK (Cheshire and Merseyside, Grampian, North West London and West Yorkshire). We examine how long different population groups were waiting for elective care, the reasons why people leave the waiting list and the health care use of people waiting for elective care, as well as how each of these vary between population groups. We conclude with some considerations for policymakers and decision makers.
The Health Foundation and the four NDL labs all undertook patient and public involvement and engagement at every stage of our analyses. These insights were central in shaping the research questions and in communicating our findings.
For the three locations in England, information on waiting times for elective care was drawn from NHS England’s Waiting List Minimum Dataset (WLMDS). The WLMDS is a patient-level collection of data relating to elective care activity, demand and waiting lists in England. It has been collected from providers of NHS services that fall within the scope of referral to treatment (RTT) requirements on a weekly basis since April 2021. For more information on our approach to processing the WLMDS, see our related blog.
In Grampian, Scotland, information on waiting times was drawn from the Scottish Morbidity Record’s General Acute Inpatient and Day Case dataset, managed by Public Health Scotland.
In all NDL locations, data on waiting times were then linked to local datasets covering patient-level demographic and health characteristics; secondary and primary care use; and prescriptions (see our technical appendix for more). Some supporting statistics on emergency care use surrounding elective procedures in England were drawn from the patient-level Hospital Episode Statistics dataset.
Data on costs for health care activity were drawn from a variety of sources. In England, patient-level costing based on Healthcare Resource Groups (HRGs) were available for secondary care categories. However, costing emergency inpatient admissions was not possible in the English NDL labs due to extensive missing data. GP appointments in England were drawn from the Personal Social Services Research Unit (PSSRU) Unit Costs for Health and Social Care publication. Grampian used average costs provided by NHS Grampian for hospital-based care.
For the NDL labs in England, the study population for this analysis was all patients referred to the NHS elective care waiting list in an NDL location between 1 April 2022 and 31 March 2024. In NDL Grampian, the study population was all NHS Grampian patients who received an elective procedure in 2022 or 2023. We also collected data for each patient covering an additional pre-referral period of 3 months and a follow-up period until 6 months after the date of latest pathway ending or 31 October 2024, whichever was sooner.
In England, waiting times for this study were defined according to RTT rules, where a wait begins with a patient’s initial referral to an elective care pathway. In Scotland, waits were defined as inpatient waits in line with the Treatment Time Guarantee standard, where waits begin with agreement for inpatient or day case treatment (for more on our definitions and methods, see our technical appendix).
For all completed waiting list pathways in NDL localities based in England, we calculated mean and median wait times in days. We then analysed the reasons people left the waiting list, breaking down the cohort of completed pathways by the stated reason for their ‘clock stop’ in the WLMDS. A clock stop indicates that a waiting period has finished, either due to receipt of a first treatment, the initiation of active monitoring, a decision not to treat by either the patient or clinician, failure to attend an appointment or the patient’s death.
NDL labs, for completed pathways with a definitive treatment and patient-level linkage to local electronic health records, recorded health care use and costs across three time periods: the 3 months prior to referral; across the total waiting period; and the 3 months following treatment. We recorded health care use in the form of GP appointments, A&E attendances, emergency admissions and elective admissions, and the number of days on which prescriptions were issued for any medications, pain relief, antibiotics or antidepressants. Throughout this briefing, health care use metrics are reported in terms of incidents per 100 people per week.
Each of the above metrics was then broken down by the following demographic and health characteristics, pertaining to the nature of the pathway or characteristics of the patient on that pathway:
- specialty
- sex
- age band
- ethnic category (ONS categories)
- Index Of Multiple Deprivation (IMD) quintile – national IMD with all components
- number of long-term conditions – split into no conditions; one condition; comorbidities; and multimorbidities
- Electronic Frailty Index categories (where available).
Assessment of the wider consequences of waiting – including impacts on people’s quality of life, mental health, progression of comorbidities and economic wellbeing – were out of scope for this work. In most cases it would not have been possible to fully address these across all NDL labs with the data we had access to. However, we would recommend that future research address these by incorporating a wider range of data (such as patient-reported outcome measures) or through qualitative investigation.
Key variables of the Waiting List Minimum Dataset, most notably the variable detailing reasons for leaving the waiting list, were not useable in some NDL labs. This ultimately limited our analysis of reasons for leaving the waiting list to one NDL location, restricting our ability to compare these metrics between geographies and draw more general conclusions. Sample sizes for our analysis of excess health care use were also heavily restricted by data quality in English labs, as only a small proportion of pathways in the WLMDS could successfully be linked to secondary care datasets (for more see our technical appendix and related blog).
Figure 2.1
Figure 2.2
People living in more deprived areas tended to experience longer wait times than those in less deprived areas (Figure 2.1 and 2.2). An exception to this could be seen in Cheshire and Merseyside, where people living in areas in the most deprived quintile had the shortest median wait time. This may relate to Cheshire and Merseyside’s approach to waiting list prioritisation, but more thorough evaluation of local prioritisation approaches would be necessary for this to be fully understood. Differences in wait times between quintiles of deprivation were less pronounced, however, when accounting for differences in age distribution. Across the three NDL locations, white patients were also seen to have shorter median wait times than patients from Asian, Black or mixed backgrounds. These differences remained after accounting for the age, frailty and deprivation distribution of the populations. However, there remains the possibility that some of these differences are due to variation in health need and the predominant specialties of waiting list pathways among different populations.
Socioeconomic and ethnic inequalities on the elective care waiting list in England have previously been observed by The Strategy Unit (2021), The King’s Fund (2023) and Nuffield Trust (2024), and have been attributed to greater barriers in attending appointments, difficulties navigating the health system for people with a first language other than English, and poorer access to care relative to local need in deprived areas.
Wide variations in waiting times were evident between specialties (Figure 3). Neurology, trauma and orthopaedics, gynaecology, general surgery and rheumatology were among the categories with the longest waits. For surgical specialties that might require overnight stays in hospital for treatment, such as trauma and orthopaedics, these extended wait times may be explained by the longer lengths of stay required by each patient, limiting capacity for treatment. For others, including gynaecology and ear, nose and throat, longer wait times might be connected to a wider range of factors, including shortages of specialists and the de-prioritisation of low mortality conditions as part of efforts to increase access to elective care following the COVID-19 pandemic.
Figure 3
Wait times differed dramatically between locations for some specialties. In North West London, for instance, the median wait for neurosurgery was more than double the length in any other NDL lab (173 days compared to 52 and 72 days in West Yorkshire and Cheshire and Merseyside, respectively). In West Yorkshire, some categories of paediatric services have median waits nearly four times those in Cheshire and Merseyside; meanwhile, West Yorkshire’s median wait time for ear, nose and throat services was only around one-quarter as long as those in the other two locations. Specialty-level differences appear to be a key driver of variation in overall median wait times – specialties where West Yorkshire has particularly low median wait times tend to be high-volume services, including ear, nose and throat (24 day median wait time, accounting for 7.1% of the region’s overall elective waiting list); respiratory medicine (27 day median wait, 6.9% of the waiting list); and gastroenterology (32 days, 6.2% of the waiting list).
Our partners in NDL West Yorkshire also analysed cases where patients were waiting for elective treatment in more than one specialty at the same time. This made up 15% of the total cohort in that locality. Patients who were only waiting for treatment in a single specialty generally experienced shorter waiting times, with 74% to 80% of waits lasting for less than 18 weeks and less than 3% waiting over a year.
In contrast, those in overlapping pathways faced significantly longer waits, with only 54% to 62% treated within 18 weeks and 5% to 9% waiting over 52 weeks. These differences were particularly pronounced in younger groups and those without long-term conditions. Almost 10% of children aged 10 years and younger in overlapping pathways waited for more than a year, compared with only 4% for those in non-overlapping pathways. This possibly indicates that overlapping waits create additional scheduling complications and bottlenecks in the health care system, leading to prolonged waits and reduced efficiency in patient care delivery.
In analysing the Waiting List Minimum Dataset (WLMDS), we found that several key fields were poorly completed by many NHS sites. This included fields recording the intended procedure and the reason for a clock stop, the latter being central to our analysis of why patients leave the waiting list. Data quality issues in these areas significantly limited our analysis.
In West Yorkshire, around 50% of waiting periods lacked a recorded reason for the clock stop; in North West London almost 40% were marked as unknown or as an unspecified non-treatment category. In contrast, this field was well completed in NDL Cheshire and Merseyside, where only 5% were unknown. Accordingly, this section reports only on analysis from that NDL lab.
For more information on the challenges we encountered working with the WLDMS and our recommendations for improvement, see our Medium blog on the topic.
Understanding why people leave the waiting list is crucial to reducing wait times, given the insights it can provide into blockages in the system and pathways ending. These results give us some insight into the variety of reasons that elective care pathways end, although they must be understood as a snapshot in one particular region. For a comprehensive understanding, further regional comparisons or analyses across time are required. We suggest further investigation into the collection practices of sites with high-quality recording of pathway ends, and the dissemination of learnings from them to other providers.
In Cheshire and Merseyside, elective care pathways finished with the receipt of a first treatment in less than half of cases (47%), while almost a third of pathways (30%) ended due to a decision not to treat. These proportions are broadly consistent with those seen on a national level in NHS England data presented at the 2025 European Operational Research conference. Around 1 in 8 pathways ended due to the initiation of active monitoring, while almost 5% of pathways ended with the patient declining treatment. Pathways ending due to death or a failure to attend were rare (0.5% and 0.4% of all pathways, respectively). The reasons people left the waiting list varied substantially by specialty. Plastic surgery had the highest rate of clock stops due to first treatment (75%), while elderly medicine frequently saw pathways end due to the initiation of active monitoring (32% of all elderly medicine pathways). For some specialties with particularly long average waiting times, including neurology and gynaecology, a relatively high proportion of pathways (over 40%) ended due to a decision not to treat. In these cases longer waits may have resulted in the patient seeking out alternative care arrangements, an outcome that would primarily fall under this category.
Figure 4
While rare overall, the proportion of pathways that ended because the patient did not attend their appointment (DNAs) varied by age, deprivation and ethnicity. People living in areas in the most deprived quintile had rates of pathways ending due to non-attendance more than double those living in the least deprived quintile (0.5% of pathways compared with 0.2%). Much like observed variations in wait times, these differences highlight barriers in access to elective care that make attendance more difficult for some population groups, as observed in previous research. These may include additional difficulties taking time off work, caring responsibilities, transport issues or obstacles to communication. However, we do not know to what extent differences in age profiles or specialty mixes between population groups may also contribute to variation in levels of DNAs.
Across all NDL locations, more deprived quintiles had higher rates of A&E attendances and emergency admissions than less deprived quintiles before, during and after waiting. In all cases, the most deprived quintile had a rate of A&E attendances between one and a half times and double the rate of the least deprived quintile in all time periods. These disparities were particularly pronounced while waiting for care: patients living in the most deprived quintile had rates of A&E attendances while waiting between 64% and 87% higher than the least deprived quintile, a greater percentage difference than in any other period (Figure 6).
Figure 6
This trend was seen less consistently for other types of health care use – while in West Yorkshire and Cheshire and Merseyside, more deprived quintiles had marginally higher rates of GP appointments compared with less deprived quintiles; in North West London less deprived quintiles tended to have the highest rates of primary care use. However, our analysis only covers NHS-funded care. As such, we cannot tell the extent to which individuals receive support from private health care, over-the-counter medication or other alternative sources of treatment.
We also performed analyses examining the impact of waiting for long periods of time on health care usage and costs.
These analyses compared the health care use of people who waited for the ‘target amount of time’ (18 weeks or less for a total RTT wait in England; 12 weeks or less for an inpatient wait in Scotland) to that of people who waited for longer than target periods of time. Through these comparisons, we aimed to identify excess health care use caused by longer waits (ie health care use we would not have expected to happen had the patient waited for the target amount of time or less).
This approach, an adaptation of difference-in-difference methods, compared each group’s use of health care while waiting (the ‘waiting period’) to their health care use during a ‘reference period’ of equal length, where we would expect their use to be at a ‘typical’ level. For this reference period, we selected a period equal to each patient’s wait length after they had received their awaited procedures, with the weeks immediately after treatment removed to account for immediate complications or aftercare. Our method of identifying excess was based on the assumption that the ‘normal’ level of health care use would be lower than that seen while waiting.
Our analysis of excess health care use found that long waits can significantly increase consumption of pain relief medication, as well as some evidence of excess primary care use due to long waits (albeit not statistically significant). However, for most procedures and points of delivery we examined, our method appeared to be unsuitable for calculating excess, as health care use remained high for months after leaving the waiting list and was often higher than during the wait itself (as was reflected in our analysis of health care use before, during and after waiting).
This opens up space for future research to investigate the reasons for health care use, and particularly emergency care use, remaining so high following receipt of elective procedures. Further investigation is also warranted into the effect of long wait times on health care use following treatment.
In the English NDL labs, data quality issues meant that only a small proportion of pathways in the WLMDS could successfully be linked to secondary care datasets (see our technical appendix and related blog), substantially limiting our sample sizes. In some cases, particularly in our analysis of GP appointments, this impacted our ability to detect more modest effects even when some evidence of excess health care use was apparent.
We are grateful to the members of the Networked Data Lab's waiting list patient and public panel who took the time to review and help improve this work, including Helen Parton-Frost, Linda Griffith, Jeremy Dearling, Saqia Ahmed, and Rashmi Kumar.
We would also like to thank Tim Gardner and Steve Wyatt for their contributions and comments on earlier drafts, as well as Chamut Kifetew and Zoe Ruziczka for managing the Networked Data Lab programme at the Health Foundation. We would also like to thank everyone who provided feedback on our preliminary results, as well as Suchi Collingwood and Sandra Rochfort from NHS England for their assistance with data processing for the Waiting List Minimum Dataset (WLMDS). We would also like to thank Marion Sumerfield and the NDL Grampian PPIE panel for quotes on their lived experience of waiting for elective care. This work uses data provided by patients and service users and collected by health and social care services as part of their care and support.
Konstantinos Daras
Benjamin Barr
Julia Barber
Tom Butterworth
Andy Pennington
Roberta Piroddi
Yuxuan Yang
Jessica E Butler
Frank Popham
Corri Black
Raul Berrocal-Martin
Sharon Gordon
Bernhard Scheliga
Irmina Zborowska
NDL Grampian PPIE panel
Marcus Yarwood
Jodie Chan
Kyle Lee-Crossett
Rafal Kulakowski
Roberto Fernandez Crespo
Uchenna Agu
Tom Clutterbuck
Alex Cheuk
Owen Melbourne
Melanie Leis
Matthew Chisambi
Sadia Khan
Chris Russell
NDL North West London PPIE panel
Anna Palczewska
Josh Elvidge-Murgatroyd
Alex Brownrigg
Helen Butters
Frank Wood
Sophie Blow
Chris Cooper
Barbara Coyle
Tom Daniels
Souheila Fox
Isaac Grinnell
Matthew Hird
Alex Jojic
James Lodge
Liz Micklethwaite
Eloise Pearson
Brian Smith
Klaudia Sutcliffe
Charlotte Waite
Tatum Yip
Waiting List Patients and Carers Panel