SEARCH

SEARCH BY CITATION

Keywords:

  • waiting times;
  • health policy;
  • patient-centred care;
  • NHS target policies;
  • waiting lists

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

For the last decade, stringent monitoring of waiting time performance targets provided English hospitals with incentives to reduce official waiting times for elective surgery. It is less clear whether the total amount of time patients waited in secondary care, from first referral to outpatient clinic until treatment, has also fallen. We used Hospital Episode Statistics inpatient data for patients undergoing total joint replacement during a period of active monitoring of targets (between 2006/7 and 2008/9) and linked it to outpatient data to reconstruct patients' pathway in the 3 years before surgery and provide alternative measurements of waiting times. Our findings suggest that although official waiting times decreased drastically in our study period, total waiting time in secondary care has not declined. Patients with shorter official waits spent a longer time in a ‘work-up’ period prior to inclusion in the official waiting list, and socio-economic inequities persisted in waiting times for joint replacement. We found no evidence that target policies achieved efficiency gains during our study period. Copyright © 2013 John Wiley & Sons, Ltd.

INTRODUCTION: WAITING TIME TARGETS AND POLICY

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

Since the 1990s, successive governments have put in place supply-side policies to reduce waiting times for elective procedures (Harrison and Appleby, 2009a, Willcox et al., 2007). These culminated in progressively shorter maximum waiting time targets and stricter penalties for hospitals unable to meet them, earning the label of ‘Targets and Terror’ policy (Propper et al., 2008). In the 1980s, a substantial minority of NHS patients spent more than 1 year on waiting lists for hip replacement surgery (Rajaratnam et al., 1990). Initially, targets focused on discrete portions of the waiting period (e.g. from GP referral to outpatient appointment or from placement on the waiting list to elective surgery). As a result, there were incentives to reduce these targeted portions of the waiting time compared with other segments of the patient's treatment pathway unaffected by the policy (Hanna et al., 2005, Harrison and Appleby, 2009a, Robinson et al., 2003). In light of criticisms (Harrison and Appleby, 2009b, Smith, 1994), the Department of Health (DoH) introduced the 18-week referral to treatment target in 2008 (Department of Health and 18 Week Pathway Programme, 2006). Whereas previous target policies started the count from the date of inclusion in the official waiting list, this target provides a more comprehensive measure of secondary care waiting times. The waiting time ‘clock’ starts from referral to secondary care, but complex rules on when to pause, stop and restart the clock (Department of Health, 2010a) mean that this target may still not match patient perceptions of waiting times. For example, when patients are referred back to primary care, miss an appointment, receive a first-line outpatient therapy or decline treatment, the clock is stopped, even though the patient may still have symptoms and perceive that they need surgery (Preston et al., 1999). Official figures indicate that more than 90% of patients were treated within 18 weeks by December 2008 (Department of Health, 2010b).

To date, it is unknown whether the various target policies have been effective in reducing patients' total clinical pathway in secondary care prior to elective surgery. Our study looks beyond the official waiting time (OWT) statistics and provide an alternative measure of waiting, to encompass the full wait in secondary care and gain insight into the patients' pathway for total joint replacement (TJR). We used Hospital Episode Statistics (HES) data on a cohort of patients undergoing primary total hip (THR) and knee replacement (TKR) between 2006/7 and 2008/9 and linked these patients' HES inpatient data with their outpatient data in the 3 years prior to their surgery. Primary THR and TKR were chosen as they are among the most common elective procedures performed in the English NHS accounting for 15% of non-emergency admissions in 2010/11 for the Trauma and Orthopaedics specialty (The Information Centre Hospital Episode Statistics, 2011a, 2011b). The years between 2006/7 and 2008/9 covered a period of increasingly stringent monitoring of NHS waiting times, while allowing us to obtain a 3-year period of outpatient data available in HES prior to surgery for all patients in our cohort.

We develop a measure of total waiting time (TWT) starting from the first referral to the first orthopaedics appointment in secondary care in the 3 years prior to surgery until the admission date for surgery. Our measurement of TWT includes the HES OWT, the period from inclusion in the official waiting list for joint replacement until actual surgery and a ‘work-up’ period in secondary care, where patients receive orthopaedic or other treatment prior to inclusion in the official waiting list. We aim to describe the trends of waiting times in this period of continuous active monitoring of waiting time targets. We then test whether the official and work-up portions of their waiting time decreased between 2006/7 and 2008/9 study the relationship between the number of days waited in the official and work-up portions of TWT. We also explore which patient and hospital characteristics were associated with differences in waiting times.

In the background section of the paper, we describe the trends for joint replacement surgery and discuss implications of rationing by waiting in terms of efficiency, equity in access to care and health outcomes for patients. Section 3 describes the methodology applied to obtain our sample data, our waiting time definitions and explanatory variables. We further present our models and the statistical methods used to analyse the data. In Section 4, we present the results from our analyses. Section 5 discusses and concludes the article.

BACKGROUND

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

Between 2006/7 and 2008/9, HES recorded a 17% increase (from 102 833 to 120 637) in primary TJR procedures performed in the English NHS. In the past, projections for primary hip and knee replacements for the UK (Dixon et al., 2004)and the USA (Kurtz et al., 2007) have underestimated the current provision, with more than 59 000 primary hip and 72 000 primary knee replacements performed in 20011/12 in the English NHS (The Information Centre Hospital Episode Statistics). Supply-side policies to increase the number of surgeries performed have also been in place (Lewis and Dixon, 2005, Brereton and Vasoodaven, 2010), introducing the right for patients to choose healthcare providers (Department of Health, 2004) and increasing healthcare market competition (Cooper et al., 2010) by contracting out joint replacement surgeries to the independent sector, but they may still not be sufficient to meet the demand for TJR in nations with ageing populations.

Target policies implicitly assume that Pareto efficiency gains can be achieved by reducing waiting times. But if an economy is performing close to its production possibilities frontier, lower waiting times may incur significant societal costs. Although waiting lists are an established rationing system in the English NHS, there are equity and efficiency concerns associated with them (Cullis et al., 2000, Culyer, 1989, Deacon and Sonstelie, 1989, Propper, 1995). Cullis and Jones argued that although reducing elective waiting lists may be desirable from a political standpoint, as it decreases the number of patients waiting (thus unhappy voters), it may be at the expense of a deadweight loss, as equally valuable but less visible or vocal NHS patients are neglected (Cullis and Jones, 2007). The opportunity cost of waiting may lead to wealthier patients exiting the waiting list and seeking health care in the private market (Besley et al., 1999, Fabbri and Monfardini, 2009), leaving poorer and perhaps sicker patients waiting in the NHS (Dixon-Woods et al., 2005). Research has shown that disparities in waiting times between sex, age and socio-economic groups have been declining (Ackerman et al., 2005, Cooper et al., 2009). Currently, there is neither evidence of socio-economic differences in the OWT for elective surgery in the English NHS (Dimakou et al., 2009) nor prioritisation by higher burden of disease (McHugh et al., 2008). However, Judge et al. showed that there is unmet need for hip and knee replacements in the English NHS and inequity in provision rates prevails by age, sex, deprivation, rurality and ethnicity, and that these vary by region (Judge et al., 2010).

Many patients go through the waiting process without delays, but a smaller residual can wait very long times (Dimakou et al., 2009, Laudicella et al., 2012). Patients value shorter waits (Lofvendahl et al., 2005) but, in most cases, place a higher value on the quality of their surgeons (Conner-Spady et al., 2008, Llewellyn-Thomas et al., 1998) and often defer to the surgeon's expertise in the decision to delay or expedite surgery (Gooberman-Hill et al., 2010). This is not the case for more severely afflicted and older patients (Conner-Spady et al., 2005). Overall, it is not clear what length of wait would be acceptable to patients.

It is also unclear what impact waiting has on TJR outcomes. Two randomised controlled trials in TKR and THR patients showed no evidence that early surgery (within 3 months) compared with late surgery (median wait 7 to 8 months) was associated with better quality of life at the time of surgery (Hirvonen et al., 2007, Hirvonen et al., 2009), nor did it result in better THR outcomes at 3 and 12 months post-operatively (Tuominen et al., 2009). TKR patients with short waits reached better quality of life earlier, but those having longer waits had better quality of life 1 year post-operatively (Tuominen et al., 2010). A systematic review on the impact of waiting for TJR (Hoogeboom et al., 2009) suggests that pain and physical function do not deteriorate for patients waiting less than 6 months.

It is the public's perception of the detrimental effect of waiting that puts pressure on Governments to act on waiting times. Although shorter waiting times are welcomed by patients, when resources are scarce, there may be a trade off between achieving short waiting times for elective surgery while maintaining high quality care and the capacity to treat emergency and priority cases. It is unclear what the optimal long-term equilibrium waiting time period in the English NHS should be or whether shorter waiting times produce higher welfare gains for society.

METHODS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

Data and sample

Our analyses used linked HES inpatient and outpatient datasets to retrospectively identify a cohort of patients admitted for TJR surgery between 2006/7 and 2008/9. HES inpatient data provide patient level information on the admission date for surgery, specialty of surgeon, date of inclusion on the official elective surgery waiting list, demographic and socio-economic characteristics and provider hospital characteristics. The HES outpatient data covered the first available years from 2003/4 to 2008/9 and provided information on appointment dates and referral date, source of referral to each appointment, specialty of the clinic and whether it was a first attendance or follow-up appointment. Each ‘first attendance’ may initiate a string of appointments. Patients often have several strings of orthopaedic appointments before being placed on the official waiting list for surgery. The HES outpatient data also included a field for diagnosis, but in practice, this usually contained a non-specific diagnosis code.

Patients who underwent primary TKR or THR between 2006/7 and 2008/9 were identified using the Office of Population Censuses and Surveys Classification of Interventions and Procedures (OPCS-4) procedure codes W37.1, W38.1, W39.1, W40.1, W41.1 or W42.1. We requested HES data for patients with these codes listed in the first HES procedure code field (The Information Centre Hospital Episode Statistics, 2009a), which is the main and most resource intensive procedure in their inpatient episode. This resulted in a cohort of 340 485 patients. We included only the first procedure for patients with multiple TJR operations during this period. We then used a random number generator to select a sample of 2000 patients and proceeded to link them by the pseudoanonymised HES identification number to HES outpatient data. We then applied further criteria to exclude non-elective and non-NHS patients and patients with missing or invalid data (Figure 1). We included only secondary care appointments occurring in the 3 years prior to admission to surgery so that each patient would have the same amount of secondary care data available, irrespective of year of surgery. Our final sample included 1618 patients with a total of 16 475 outpatient appointments, of which 6934 were Trauma and Orthopaedics appointments.

image

Figure 1. Patient flow diagram. HES, Hospital Episode Statistics; TJR, total joint replacement

Download figure to PowerPoint

Measuring total waiting times in secondary care: HES official and work-up waiting times

We estimated TWT as the difference in days between the date of admission for surgery and the date of referral to the earliest orthopaedics outpatient appointment in the 3 years before surgery. TWT includes two distinct periods of wait: the official period, from the date a consultant includes a patient in the official waiting list, and the ‘work-up’ period, a period of wait from first referral to orthopaedics clinic in the 3 years prior to surgery until inclusion in the official waiting list (Figure 2). This work-up period includes waiting for the first secondary care referral as well as other secondary care treatment, such as diagnostic tests, referrals for second opinions and treatment of comorbidities, which can occur prior to inclusion on the official waiting list for surgery. It is a period of overspill wait in secondary care that is not included in the OWT statistics. The HES OWT for elective surgery is calculated as the difference in days between the date of admission for surgery and the official date of decision to admit for surgery. HES OWT starts on the same date as the inpatient waiting time statistics reported by the DoH but also includes periods of ‘paused’ and ‘stopped’ clocks, excluded from DoH figures (Department of Health, 2010a), as discussed in Dixon (2004).The difference between TWT and OWT is the work-up waiting time (WWT) period. Patients may or may not experience work-up waiting periods. If the clock starts and the NHS hospital chooses to send the patient to an independent hospital, the clock keeps running. Usually, the independent sector hospital will guarantee that the patient will be treated before the breach date for the NHS hospital. In our analysis, we treated all referred patients as contributing to the waiting times of the independent hospital, not the referring NHS hospital.

image

Figure 2. Pathway for our cohort of patients awaiting total joint replacement. OWT, official waiting time; TWT, total waiting time; WWT, work-up waiting time

Download figure to PowerPoint

Explanatory variables: patients and hospital characteristics

Patients' demographic and socio-economic characteristics were age, sex and ethnicity, rural or urban residence, and a deprivation indicator based on the overall index of multiple deprivation, taking the value one if patients lived in a neighbourhood in the 20% least deprived areas and zero otherwise.

Patients' clinical characteristics were admission type (from a waiting list, booked or planned admission), joint replaced (hip or knee) and the presence of comorbidities through the computation of the Charlson comorbidity index (CCI) (Charlson et al., 1987). Patients with a weighted CCI sum of 2 or higher were grouped into a single category. In a booked admission, patients are given a date for surgery at the time the decision to operate is made that they then need to wait for. In a planned admission, they are also given a date for surgery, but usually as a planned sequence of care (e.g. left hip surgery followed by left ankle surgery)

Hospital characteristics were the size of the orthopaedics department, derived by ranking hospitals by the number of TJR procedures performed in our patient sample and using tertiles to define small, medium and large departments; type of hospital (i.e. NHS foundation trust, independent sector provider or other NHS provider); and the government office region where the hospital was located.

Model and statistical methods

All analyses were conducted using version 11 of Stata (StataCorp (2009). Stata Statistical Software: Release 11. College Station, TX: StataCorp LP.). Nine patients had a WWT of zero, so WWT was increased by 1 day for all patients to avoid the log of zero missing values as advised in the literature (Kirkwood and Sterne, 2003). The average number of days waited was estimated for OWT, WWT and TWT. We used simple ordinary least squares regression to estimate univariable associations between each potential explanatory variable and waiting time definitions.

We then used multiple regression analysis to further investigate the most significant predictors of OWT and WWT, the two mutually exclusive components of TWT. The multiple regression analyses models were as follows:

  • display math

where X is the matrix of patient socio-demographic factors (i.e. age, sex, ethnicity, urban indicator and living in the top 20% less deprived areas), W is the matrix of patient clinical factors (e.g. CCI index = 0, 1, ≥2), Z is the matrix of hospital factors (i.e. size of orthopaedics department, type of hospital provider and region), β and φ are vectors of parameters, ε and ν are vectors of error terms, and s, c and h are subscripts for the socio-demographic, clinical and hospital factors, respectively. Graphical analysis (Figure 3) and the Shapiro–Wilk (Shapiro and Wilk, 1965) and Shapiro–Francia (Shapiro and Francia, 1972) tests for normality of OWT and WWT suggested that both distributions were non-normal and positively skewed, so we estimated the models by general linear model (GLM) regression. The modified Park test (Manning and Mullahy, 2001) recommended a Poisson distributional family for both OWT and WWT. Ladder-of-powers histograms indicated that either square root or log transformations could be appropriate. The Pregibon link test (Pregibon, 1980) accepted the log transformation and rejected the square root transformation for both OWT and WWT at the 5% level. A modified Hosmer and Lemeshow goodness-of-fit test (Hosmer and Lemeshow, 2000) of the residuals on deciles of fitted values suggested that both GLM models with Poisson family and log link models were well calibrated. White's robust standard errors accounted for the heteroskedasticity found in the error term (Breusch and Pagan, 1979). Coefficients were exponentiated for ease of interpretation.

image

Figure 3. Distributions of official and work-up waiting times, in study period and by year

Download figure to PowerPoint

We further fitted a seemingly unrelated estimation (SUE) regression without constraints to account for cross-model covariance. We checked the variance–covariance matrix for OWT and WWT and tested whether each regressor's coefficient was equal across models. To check robustness of our models, we fitted a seemingly unrelated regression (SUR) model on log transformed OWT and WWT with robust standard errors. We also computed the correlation between the matrix of residuals and performed the Breusch–Pagan test of independence of residuals (Breusch and Pagan, 1980) to test whether OWT and WWT were correlated.

RESULTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

Sample, descriptive and simple regression analyses

After exclusions, there were complete data for 1618 of 2000 individuals sampled (942 TKR and 676 THR operations) who had 16 475 appointments, of which 6934 were orthopaedic appointments (Figure 1). Patients in our cohort were on average 69 years old (SD 10.0), predominantly British White (80%) and female (57%, Table 1). The CCI ranged from 0 to 9 in our cohort, with 25% of patients having a positive CCI (>0), but only 6% had a score of 2 or above. Twenty-eight per cent and 24% of patients treated in an NHS foundation or other trust had a positive CCI compared with 6% in the independent sector. Our cohort was similar, in terms of mean age and gender, to all patients undergoing TJR in this period (National Joint Registry for England and Wales, 2011).

Table 1. Descriptive statistics for our cohort of NHS patients by type of joint replacement
 TKA (n = 942)THA (n = 676)
Mean age (SD)70(9.0)68(11.3)
Number of males (%)403(43)290(43)
Number of British White patients (%)735(78)563(83)
Number of patients living in less deprived neighbourhoods (%)188(20)143(21)
Number of patients with positive Charlson index [CCI >0] (%)248(26)154(23)
Number of patients treated in NHS foundation trusts (%)322(34)252(37)
Of which, number with CCI > 0 (%)90(28)72(29)
Number of patients treated in the independent sector (%)34(4)29(4)
Of which, number with CCI > 0 (%)4(12)0(0)
Number of patients treated in other NHS trusts (%)586(62)395(58)
Of which, number with CCI > 0 (%)154(26)82(21)
Mean number of appointments (SD)10.6(8.3)9.6(8.7)
Mean number of orthopaedics appointments (SD)4.5(3.1)3.9(2.9)
Mean number of strings of appointments (SD)1.4(0.8)1.5(0.9)

Patients had on average 10.2 appointments (SD 8.5) in the 3 years before surgery of which 4.3 (SD 3.1) were in orthopaedic clinics. The distribution of appointments is positively skewed, whereby most patients attended fewer appointments (median 8.0 for total and 3.0 for orthopaedic appointments) and a minority (5%) attended more than three times that number, over 27 and 10 appointments, respectively. The most frequent types of non-orthopaedic appointments were ophthalmology (11.2%), rheumatology (8.4%), general surgery (7.6%) and cardiology (5.9%). GPs made 76% of the referrals to the first orthopaedics appointment in the 3-year period, 10% were made by other secondary care consultants and the remaining 14% had other or unspecified sources of referral. Eighty-two per cent of these initial orthopaedics appointments were ‘first attendances’, and 56% were led by the Trauma and Orthopaedics consultant who subsequently performed the TJR surgery. Thirty-six per cent of the patients had at least one string of orthopaedic appointments prior to the string that led to inclusion in the official waiting list. Patients who had TJR in 2006/7 had an average of 3.9 orthopaedic appointments (SD 2.8) in the 3 years prior to surgery, compared with 4.4 (SD 3.3) for patients in 2008/9. This is a 13% increase (p = 0.01) in the average number of orthopaedic appointments, compared with a 9% increase in all outpatient appointments, from average 9.6 (SD 8.1) to 10.5 (SD 8.7) appointments, in the same period (p = 0.1).

Official waiting time fell from a mean of 157 days in 2006/7 to 88 days in 2008/9 (44% reduction; p < 0.001, Table 2). However, there was no evidence that TWT decreased between 2006/7 (585 days) and 2008/9 (575 days; p = 0.61) and an indication that WWT actually increased between 2006/7 (429 days) and 2008/9 (487 days; p = 0.07) after an initial spike of 532 in 2007/8. There was no evidence that OWT differed by age, sex, ethnicity, deprivation rank, urban residence or joint replaced, but it was shorter for NHS patients whose surgery took place in independent sector hospitals (p < 0.001). On the other hand, there was some evidence that TWT and WWT were associated with deprivation, whereby NHS patients living in less deprived areas had at least 10% shorter waits (p = 0.05, p = 0.04), and with type of surgery, with TKR patients having around 20% longer waits than THR patients (p < 0.001). A higher proportion of NHS patients from less deprived neighbourhoods were treated in independent sector hospitals (27% vs 16% at NHS Foundations trusts and 23% at other NHS trusts, chi-squared test p = 0.001—results not in the table).

Table 2. Mean waiting times in days: official, work-up and total waiting times
 OWTWWTTWT
  • IMD, index of multiple deprivation; TKR, total knee replacement; THR, total hip replacement.

  • *

    Last row per category reports the t-test p-value for the OLS regression of waiting time definition on each variable.

By year   
2006/7157429585
2007/8119532651
2008/988487575
p-value*(0.000)(0.074)(0.610)
By age (years)   
Under 65120515635
Over 65119470589
p-value*(0.759)(0.094)(0.085)
By sex   
Male119499618
Female119464583
p-value*(0.947)(0.159)(0.163)
By ethnicity   
British White118490608
Other125459584
p-value*(0.165)(0.317)(0.439)
By IMD rank   
20% less deprived121435555
Others119496615
p-value*(0.677)(0.042)(0.050)
By urban index   
Urban122472593
Rural118487606
p-value*(0.484)(0.588)(0.669)
By admission method   
Waiting list124489613
Booked104495599
Planned111338449
p-value*(0.000)(0.101)(0.027)
By joint replaced   
TKR122527650
THR115423538
p-value*(0.066)(0.000)(0.000)
By Charlson index   
Charlson index = 0118469587
Charlson index = 1126553679
Charlson index = 2117447563
p-value*(0.418)(0.221)(0.177)
By size of orthopaedics department   
Small117489606
Medium114465580
Large126498624
p-value*(0.049)(0.736)(0.513)
By hospital type   
NHS foundation trust107486593
Independent sector63364427
Other NHS provider130490620
p-value*(0.000)(0.816)(0.237)
By hospital region   
North East113488601
North West109500608
Yorkshire98519617
East Midlands114499613
West Midlands120475595
East of England130457586
London123533657
South East131512643
South West124390514
p-value*(0.000)(0.159)(0.448)
Average mean wait119484603
Average median wait108336462

Multiple regression analyses results

After adjustment for other covariates, a typical patient (see Table 3 footnote) undergoing knee replacement in 2006/7 waited a median number of 503 days in secondary care before being included in the official waiting list and then a further 158 days on the official waiting list prior to surgery (Table 3). By 2007/8, the OWT decreased by 35 and 65 days (a 42% decrease, p < 0.001) by 2008/9. However, the secondary care WWT increased by 98 days in 2007/8 (p = 0.001) and 52 days by 2008/9 (p = 0.075) in the period. NHS patients who had surgery in independent sector hospitals spent 44% shorter waits (50 days less, p = 0.006) on the official waiting list compared with patients who had surgery in an NHS foundation trust, whereas patients in other NHS trusts waited 12% longer (13 more days, p = 0.001). Patients with booked admissions also had shorter OWTs (p = 0.003). There was no evidence of socio-economic or demographic disparities for OWT and little evidence of regional disparities in OWT. However, our results show that older patients (p = 0.011) and patients living in less deprived neighbourhoods (p = 0.039) waited fewer days in the work-up period. There is also strong evidence that patients with one comorbidity waited 84 more days in WWT than those without comorbidities (18% more, p = 0.007) and patients undergoing hip replacement surgery waited 20% less than knee replacement patients (110 fewer days on WWT, p < 0.001). There is no evidence of differential work-up times for NHS patients undergoing surgery in independent sector hospitals or other NHS trusts compared with those operated on in NHS Foundation Trust hospitals, nor evidence of regional disparities in WWT. This finding suggests that the effect of independent sector hospital on waiting times only occurs after the patient is booked onto the official waiting list. All of our estimates were robust to model specification, with OLS models on log transformed OWT and WWT producing similar estimates and standard errors.

Table 3. Multiple regression results
 GLM Poisson family with log linkSUE test p-value
OWT [exp(b)]95% CIMarginal effect (days)WWT [exp(b)]95% CIMarginal effect (days)
  • GLM, general linear model; IMD, index of multiple deprivation; SUE, seemingly unrelated estimation; OWT, official waiting time; WWT, work-up waiting time.

  • Our baseline patient is a British White male, 69 years old from a rural area, without comorbidities picked up by the Charlson comorbidity index, undergoing total knee replacement in 2006/7 and admitted from a waiting list to a NHS Foundation Trust hospital with a small orthopaedics department in the North East region.

  • *

    p < 0.10.

  • **

    p < 0.05.

  • ***

    p < 0.01.

Year: 2006/7Reference  Reference   
2007/80.773***[0.720–0.829]−351.226***[1.081–1.389]980.000
2008/90.579***[0.539–0.622]−651.120*[0.989–1.269]520.000
Age (mean centred)0.998[0.995–1.001]00.994**[0.989–0.999]−30.201
Male0.986[0.929–1.046]−20.931[0.845–1.027]−350.342
Non-British White1.021[0.945–1.103]20.914[0.802–1.043]−430.169
IMD: 20% less deprived1.016[0.939–1.098]20.872**[0.766–0.993]−660.055
Urban area0.966[0.898–1.040]−40.963[0.853–1.088]−180.964
Admission: waiting listReference  Reference   
Booked admission0.887***[0.820–0.960]−140.996[0.888–1.117]−20.118
Planned admission1.250[0.904–1.730]300.732*[0.507–1.057]−1320.044
Hip joint replacement0.953[0.897–1.013]−60.797***[0.720–0.882]−1100.004
Charlson index = 0Reference  Reference   
Charlson index = 11.074*[0.987–1.168]91.179***[1.047–1.328]840.224
Charlson index ≥ 20.982[0.846–1.139]−20.996[0.797–1.245]−20.919
Size of orthopaedics department: smallReference  Reference   
Medium0.958[0.887–1.035]−50.983[0.868–1.112]−80.741
Large1.040[0.961–1.125]51.031[0.906–1.174]150.918
Hospital type: NHS FoundationReference  Reference   
Independent sector0.556***[0.365–0.846]−500.990[0.668–1.469]−50.064
Other NHS Trust1.118***[1.046–1.194]130.998[0.887–1.123]−10.110
Region: North EastReference  Reference   
North West0.891[0.759–1.046]−131.069[0.829–1.378]330.241
Yorkshire0.821**[0.696–0.969]−221.064[0.814–1.390]300.101
East Midlands0.949[0.801–1.126]−61.031[0.778–1.367]150.627
West Midlands0.960[0.829–1.112]−50.966[0.757–1.232]−160.968
East of England1.095[0.935–1.282]120.952[0.723–1.254]−230.399
London0.940[0.801–1.104]−71.131[0.854–1.497]620.273
South East1.013[0.868–1.183]21.111[0.857–1.440]530.554
South West0.977[0.840–1.137]−30.824[0.637–1.066]−840.262
Constant159***[137–185] 504***[386–658] 0.000
No. of observation1618  1618   

Seemingly unrelated estimation tests showed that coefficient estimates for year, planned admission and undergoing a hip replacement are different between the OWT and WWT models (Table 3, last column). There is also weaker evidence (p < 0.1) that our result for NHS patients living in less deprived areas or undergoing TJR in the independent sector is also different between both models. The covariance matrix of coefficients for the SUE model displayed negative variances between year variables in OWT and WWT, indicating that by year, OWT and WWT are negatively correlated (not reported). This is substantiated by our SUR estimation results (not reported), with a correlation of residuals of −0.053 (p = 0.0342), which confirms that the models are not independent but negatively correlated; therefore, patients with shorter OWT wait longer in the WWT period.

DISCUSSION AND CONCLUSIONS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

Whereas OWTs for elective surgery have decreased drastically over the years, TWT spent in outpatient secondary care was more stable. Shorter waiting times after inclusion in the official waiting list were associated with longer waits in secondary care prior to inclusion in the official waiting list. Patients without comorbidity and those from least deprived areas spent less time in the work-up period. NHS patients who had surgery in independent sector hospitals had much shorter OWTs but did not have shorter work-up periods.

The novelty of our study was to link HES inpatient and outpatient data. To our knowledge, we are the first to link these datasets to study the secondary care clinical pathway for patients awaiting surgery. This methodology can be applied to study other clinical issues that require a holistic evaluation of pathways to inpatient health services. Our study provides a descriptive analysis of the waiting time trends in a period of active monitoring of waiting time targets. As policies evolved during the period rather than changed drastically at one point in time, we could not use analytic techniques such as difference-in-differences analysis for policy evaluation (Propper et al., 2008). The strength of analysing routine HES data is that it is comprehensive and generalisable to all patients undergoing TJR in England. However, it has some limitations. One limitation is the failure of hospitals to record a diagnosis code for outpatient care. This field is not mandatory and does not affect hospital payments, and thus was poorly recorded in our sample (98% of appointments had a non-specific code) and nationally (The Information Centre Hospital Episode Statistics, 2010). This raises three problems. First, we could not be certain whether any given orthopaedic appointment was related to the hip or knee problem subsequently treated surgically, potentially leading to an overestimate of TWT for patients with multiple orthopaedic problems. Second, it is impossible to estimate how many patients wait for longer than 3 years in secondary care for treatment of their orthopaedic problem, which could lead to an underestimate of TWT in some patients (Armstrong, 2000, Sobolev et al., 2001). And third, we do not know which interventions were used in outpatient care, in an attempt to delay the need for surgery. Despite these measurement problems, if the patient cohorts undergoing TJR during successive years of our study period are similar in terms of sociodemographics and clinical factors, then our estimates of the trends in work-up and OWTs will still be valid. The exception being if a significant number of outpatient appointments were not captured in the first years of HES outpatient data that we use. In aggregate, early HES data achieved close to 100% coverage when compared with pre-existing outpatient activity statistics (The Information Centre Hospital Episode Statistics, 2006), leading us to believe that this is a very minor issue for our cohort. Finally, our results may be affected by unmeasured confounding factors. For example, there may be unmeasured hospital and patient characteristics, such as number of hospital staff and patients' risk factors (i.e. smoking and body mass index), which may affect waiting times but are not captured in HES data.

It is unclear from our results whether waiting list initiatives are equitable or achieve efficiency gains in the delivery of TJR surgery (Cullis et al., 2000). They provide distorted incentives to reduce waiting in those patients where it is easiest and cheapest to do so. We found that a higher proportion of patients from least deprived areas move quickly onto the official waiting lists (Browne et al., 2008) and less complex cases are taken on by independent sector hospitals, whereas patients with more complex needs or from more deprived areas wait longer during work-up and on waiting lists for treatment at NHS hospitals. In a recent study of HES inpatient data, Siciliani and colleagues found that independent sector hospitals achieved shorter lengths of stay for patients undergoing hip replacement surgery, even after controlling for comorbidities (Siciliani et al., 2012). However, they also did not control for lifestyle risk factors such as smoking and obesity, and therefore, their results may still be affected by residual confounding. Further, independent sector hospitals impose patient eligibility criteria and leave the residual less healthy patient group to be treated in NHS facilities (Timmins, 2005b). Competition from the independent sector could in itself result in improved technical efficiency of care (Timmins, 2005a), but cost-effectiveness data and stronger study designs are needed (Bardsley and Dixon, 2011, Chard et al., 2011).

Our results not only add to the body of evidence that targets to reduce OWTs (Dimakou et al., 2009, Propper et al., 2008) but also suggest that the work-up waiting period before being placed on the surgical waiting list may act as a buffer for hospitals to manage patients and resources when providing elective surgery services. Although the increase in work-up time observed could represent additional care at early stages of the conditions leading to TJR, the lack of decline in TWT despite a 17% increase in supply of TJR procedures during the same period may indicate that this TWT could be closer to the long run waiting time equilibrium in the English NHS. When the performance on the 18-week referral to treatment target was temporarily unmonitored (Department of Health and NHS Finance Performance & Operations, 2010), the DoH official median waiting times increased from median 7.7 weeks in March 2009 to 9.1 weeks in January 2011 (Department of Health, 2011). More recently, the UK Government announced its commitment to the target (Ham, 2011, Secretary of State for Health, 2011) enshrined in the NHS constitution, but there are concerns that pressures to meet waiting times and financial targets come at the expense of quality of care. Knowing that surgery will be performed within 18 weeks may well provide reassurance to patients being placed on the official waiting list, but it is important not to overlook other patients, not yet on the official list, waiting for referral to a surgeon, receiving conservative care to delay the need for surgery or being treated for comorbidities. Our study highlights the difficulty of monitoring waiting time targets for surgery when there is no clear point when the patient is ready for surgery at which to start the clock. Future guidelines should focus on ensuring appropriate referral pathways from primary to secondary care, trials of conservative care where appropriate, rapid diagnostic testing and treatment of comorbidities followed by speedy surgical treatment when thresholds for surgery are met.

In conclusion, despite decreases in official waits, there is no evidence that target policies between 2006/7 and 2008/9 reduced the total amount of time waiting from referral to secondary care until surgery. Patients with shorter official waits spent a longer time in a ‘work-up’ period prior to inclusion in the official waiting list, and socio-economic inequities persisted in waiting times for joint replacement.

CONFLICT OF INTEREST

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

All authors declare no conflict of interest. All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author).

ACKNOWLEDGEMENTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES

We thank Pete Shiarly for his contribution in managing raw HES data, linking inpatient and outpatient data and generating a random sample of patients for analysis. We thank Roy Maxwell for facilitating the liaison with HES Information Centre. We thank the RESTORE team and anonymous reviewers for helpful comments.

Funding: This article outlines independent research commissioned by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research funding scheme (RP-PG-0407-10070). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. This work was also funded by the MRC Hubs for Trial Methodology Research, Grant number G0800800.

The University of Bristol is the study sponsor and ensured data were physically and electronically kept safe. The University of Bristol and the study team complied with HES Information Centre requirements to receive HES inpatient and outpatient data. The HES data extract contained only anonymised data without sensitive field codes. Ethical approval was not required.

This article followed the STROBE guidelines for observational studies. Our study explored time trends in waiting periods. We did not pre-specify study hypotheses.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION: WAITING TIME TARGETS AND POLICY
  4. BACKGROUND
  5. METHODS
  6. RESULTS
  7. DISCUSSION AND CONCLUSIONS
  8. CONFLICT OF INTEREST
  9. ACKNOWLEDGEMENTS
  10. REFERENCES
  • Ackerman IN, Graves SE, Wicks IP, Bennell KL, Osborne RH. 2005. Quantifying social disparities in orthopaedic waiting lists. Internal Medicine Journal 35: A100.
  • Armstrong PW. 2000. First steps in analysing NHS waiting times: avoiding the ‘stationary and closed population’ fallacy. Statistics in Medicine 19: 20372051.
  • Bardsley M, Dixon J. 2011. Quality of care in independent sector treatment centres. British Medical Journal 343.
  • Besley T, Hall J, Preston I. 1999. The demand for private health insurance: do waiting lists matter? Journal of Public Economics 72: 155181.
  • Brereton L, Vasoodaven V. 2010. The impact of the NHS market: an overview of the literature.
  • Breusch TS, Pagan AR. 1979. Simple test for heteroscedasticity and random coefficient variation. Econometrica, 47: 12871294.
  • Breusch TS, Pagan AR. 1980. The LM test and its application to model specification in econometrics. Review of Economic Studies 47: 239254.
  • Browne J, Jamieson L, et al. 2008. Case-mix & patients' reports of outcome in Independent Sector Treatment Centres: Comparison with NHS providers. BMC Health Services Research 8: 78.
  • Chard J, Kuczawski M, Black N, Meulen JVD. 2011. Outcomes of elective surgery undertaken in independent sector treatment centres and NHS providers in England: audit of patient outcomes in surgery. British Medical Journal 343.
  • Charlson ME, Pompei P, Ales KL, Mackenzie CR. 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. Journal of Chronic Diseases 40: 373383.
  • Conner-Spady B, Estey A, Arnett G, Ness K, Mcgurran J, Bear R, Noseworthy T. 2005. Determinants of patient and surgeon perspectives on maximum acceptable waiting times for hip and knee arthroplasty. Journal of Health Services Research & Policy 10: 8490.
  • Conner-Spady B, Sanmartin C, Johnston G, Mcgurran J, Kehler M, Noseworthy T. 2008. Willingness of patients to change surgeons for a shorter waiting time for joint arthroplasty. Canadian Medical Association Journal, 179: 327332.
  • Cooper Z, Gibbons S, Jones S, Mcguire A. 2010. Does hospital competition save lives? Evidence from the English NHS patient choice reforms. LSE Health working papers 16/2010.
  • Cooper Z N, Mcguire A, Jones S, LE Grand J. 2009. Equity, waiting times, and NHS reforms: retrospective study. British Medical Journal, 339: 673675.
  • Cullis J, Jones P. 2007. The welfare economics of public sector targets: towards an ‘expenditure constitution’. Economic Issues, 12: 114.
  • Cullis J G, Jones P R, Propper C. 2000. Chapter 23 Waiting lists and medical treatment: analysis and policies. In: Anthony J. C. & Joseph P. N. (eds.) Handbook of Health Economics. Elsevier.
  • Culyer A J. 1989. The normative economics of health care finance and provision. Oxford Review of Economic Policy, 5: 3458.
  • Deacon R T, Sonstelie J O N. 1989. The welfare costs of rationing by waiting. Economic Inquiry, 27: 179196.
  • Department of Health. 2004. “Choose and book”: patient's choice of hospital and booked appointment—policy framework for choice and booking at the point of referral.
  • Department of Health. 2010a. The 18-week rules suite—national clock rules. http://www.dh.gov.uk/prod_consum_dh/groups/dh_digitalassets/@dh/@en/@ps/@sta/@perf/documents/digitalasset/dh_113628.pdf, Accessed April 3rd 2011.
  • Department of Health. 2010b. NHS referral to treatment (RTT) waiting times statistics for England, 2009 annual report.
  • Department of Health. 2011. Statistical press notice NHS referral to treatment (RTT) waiting times data January 2011.
  • Department of Health & 18 Week Pathway Programme. 2006. Tackling hospital waiting: the 18 week patient pathway. An implementation framework.
  • Department of Health & NHS Finance Performance & Operations. 2010. Revision to the operating framework for the NHS in England 2010/11.
  • Dimakou S, Parkin D, Devlin N, Appleby J. 2009. Identifying the impact of government targets on waiting times in the NHS. Health Care Management Science, 12: 110.
  • Dixon-Woods M, Kirk D, Agarwal S, Annandale E, Arthur T, Harvey J, Hsu R, Katbamna S, Olsen R, Smith L, Riley R, Sutton A. 2005. Vulnerable groups and access to health care: a critical interpretive review. London: NCCSDO, p. 208235.
  • Dixon S. 2004. Trends in waiting time to date and total time waited: are the sources compatible? Health Statistics Quarterly, 2329.
  • Dixon T, Shaw M, Ebrahim S, Dieppe P. 2004. Trends in hip and knee joint replacement: socioeconomic inequalities and projections of need. Annals of the Rheumatic Diseases, 63: 825830.
  • Fabbri D, Monfardini C. 2009. Rationing the public provision of healthcare in the presence of private supplements: evidence from the Italian NHS. Journal of Health Economics, 28: 290304.
  • Gooberman-Hill R, Sansom A, Sanders C M, Dieppe P A, Horwood J, Learmonth I D, Williams S, Donovan J L. 2010. Unstated factors in orthopaedic decision-making: a qualitative study. BMC Musculoskelet Disorders, 11: 213.
  • Ham C. 2011. The prime minister's commitments on the NHS. British Medical Journal
  • Hanna S J, Muneer A, Khalil K H. 2005. The 2-week wait for suspected cancer: time for a rethink? International Journal of Clinical Practice 59: 13341339.
  • Harrison A. Appleby J. 2009a. Reducing waiting times for hospital treatment: lessons from the English NHS. Journal of Health Services Research & Policy, 14: 168173.
  • Harrison A J, Appleby J. 2009b. English NHS waiting times: what next? Journal of the Royal Society of Medicine, 102: 260264.
  • Hirvonen J, Blom M, Tuominen U, Seitsalo S, Lehto M, Paavolainen P, Hietaniemi K, Rissanen P, Sintonen H. 2007. Evaluating waiting time effect on health outcomes at admission: a prospective randomized study on patients with osteoarthritis of the knee joint. Journal of Evaluation in Clinical Practice 13: 728733.
  • Hirvonen J, Tuominen U, Seitsalo S, Lehto M, Paavolainen P, Hietaniemi K, Rissanen P, Sintonen H, Blom M. 2009. The effect of waiting time on health-related quality of life, pain, and physical function in patients awaiting primary total hip replacement: a randomized controlled trial. Value in Health, 12: 942947.
  • Hoogeboom T J, Van Den Ende C H, Van Der Sluis G, Elings J, Dronkers J J, Aiken A B, Van Meeteren N L. 2009. The impact of waiting for total joint replacement on pain and functional status: a systematic review. Osteoarthritis and Cartilage, 17: 14201427.
  • Hosmer D W, Lemeshow S. 2000. Applied Logistic Regression. Chapter 5—Assessing the Fit of the Model, New York: Wiley and Sons.
  • Judge A, Welton N J, Sandhu J, Ben-Shlomo Y. 2010. Equity in access to total joint replacement of the hip and knee in England: cross sectional study. British Medical Journal, 341: c4092.
  • Kirkwood B, Sterne J. 2003. Essentials of Medical Statistics 2nd edn, Wiley-Blackwell.
  • Kurtz S, Ong K, Lau E, Mowat F, Halpern M. 2007. Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. The Journal of Bone and Joint Surgery. American Volume, 89: 780785.
  • Laudicella M, Siciliani L, Cookson R. 2012. Waiting times and socioeconomic status: evidence from England. Social Science & Medicine, 74: 13311341.
  • Lewis R, Dixon J. 2005. The future of primary care: meeting the challenges of the new NHS market.
  • Llewellyn-Thomas H A, Arshinoff R, Bell M, Williams J I, Naylor C D. 1998. In the queue for total joint replacement: patients' perspectives on waiting times. Journal of Evaluation in Clinical Practice 4: 6374.
    Direct Link:
  • Lofvendahl S, Eckerlund I, Hansagi H, Malmqvist B, Resch S, Hanning M. 2005. Waiting for orthopaedic surgery: factors associated with waiting times and patients' opinion. International Journal for Quality in Health Care, 17: 133140.
  • Manning W G, Mullahy J. 2001. Estimating log models: to transform or not to transform? Journal of Health Economics, 20: 461494.
  • Mchugh G A, Campbell M, Silman A J, Kay P R, Luker K A. 2008. Patients waiting for a hip or knee joint replacement: is there any prioritization for surgery? Journal of Evaluation in Clinical Practice 14: 361367.
  • National Joint Registry for England and Wales. 2011. 8th annual report.
  • Pregibon D. 1980. Goodness of link tests for generalized linear models. Journal of the Royal Statistical Society. Series C (Applied Statistics), 29: 1514.
  • Preston C, Cheater F, Baker R, Hearnshaw H. 1999. Left in limbo: patients' views on care across the primary/secondary interface. Quality in Health Care, 8: 1621.
  • Propper C. 1995. The disutility of time spent on the United Kingdom's National Health Service waiting lists. The Journal of Human Resources, 30: 677700.
  • Propper C, Sutton M, Whitnall C, Windmeijer F. 2008. Did ‘targets and terror’ reduce waiting times in England for hospital care? The B.E. Journal of Economic Analysis & Policy, 8(2): 5.
  • Rajaratnam G, Black N A, Dalziel M. 1990. Total hip replacements in the National Health Service: is need being met? Journal of Public Health Medicine, 12: 5659.
  • Robinson D, Bell C M, Moller H, Basnett I. 2003. Effect of the UK government's 2-week target on waiting times in women with breast cancer in southeast England. British Journal of Cancer, 89: 492496.
  • Secretary of State for Health. 2011. Government response to the NHS Future Forum report.
  • Shapiro S S, Francia R S. 1972. An approximate analysis of variance test for normality. Journal of the American Statistical Association, 67: 215216.
  • Shapiro S S, Wilk M B. 1965. An analysis of variance test for normality (complete samples). Biometrika, 52: 591611.
  • Siciliani L, Sivey P, Street A. 2012. Differences in length of stay for hip replacement between public hospitals, specialised treatment centres and private providers: selection or efficiency? Health Economics.
  • Smith T. 1994. Waiting times: monitoring the total postreferral wait. British Medical Journal, 309: 593596.
  • Sobolev B, Brown P, Zelt D. 2001. Potential for bias in waiting time studies: events between enrolment and admission. Journal of Epidemiology and Community Health, 55: 891894.
  • The Information Centre Hospital Episode Statics. 2006. Outpatient data quality report 2003–04 and 2004–05.
  • The Information Centre Hospital Episode Statics. 2009. Main procedures and interventions: 4 character table 2008-9.
  • The Information Centre Hospital Episode Statics. 2010. Main procedures and interventions: 4 character table 2009–10.
  • The Information Centre Hospital Episode Statics. 2011a. Main procedures and interventions: 4 character table 2011–12.
  • The Information Centre Hospital Episode Statics. 2011b. Main specialty 2010–11.
  • Timmins N. 2005a. The NHS revolution: health care in the market place: challenges of private provision in the NHS. BMJ: British Medical Journal, 331: 11931195.
  • Timmins N. 2005b. Use of private health care in the NHS. British Medical Journal, 331: 11411142.
  • Tuominen U, Sintonen H, Hirvonen J, Seitsalo S, Paavolainen P, Lehto M, Hietaniemi K, Blom M. 2009. The effect of waiting time on health and quality of life outcomes and costs of medication in hip replacement patients: a randomized clinical trial. Osteoarthritis and Cartilage., 17: 11441150.
  • Tuominen U, Sintonen H, Hirvonen J, Seitsalo S, Paavolainen P, Lehto M, Hietaniemi K, Blom M. 2010. Is longer waiting time for total knee replacement associated with health outcomes and medication costs? Randomized clinical trial. Value in Health, 13: 9981004.
  • Willcox S, Seddon M, Dunn S, Edwards R T, Pearse J, Tu J V. 2007. Measuring and reducing waiting times: a cross-national comparison of strategies. Health affairs (Project Hope), 26: 10781087.