Are randomised trials valid externally? A comparison of patient characteristics and outcomes in an observational study and a randomised trial of acupuncture in patients with migraine

ObjectiveThere has been concern that an increase in billing for high-intensity emergency care is due to changes in coding practices facilitated by electronic health records. We sought to characterise the trends in billing for high-intensity emergency care among Medicare beneficiaries and to examine the degree to which trends in high-intensity billing are explained by changes in patient characteristics and services provided in the emergency department (ED).Design, setting and participantsObservational study using traditional Medicare claims to identify ED visits at non-federal acute care hospitals for elderly beneficiaries in 2006, 2009 and 2012.Outcomes measuresBilling intensity was defined by emergency physician evaluation and management (E&M) codes. We tested for overall trends in high-intensity billing (E&M codes 99285, 99291 and 99292) and in services provided over time using linear regression models, adjusting for patient characteristics. Additionally, we tested for time trends in rates of admission to the hospital and to the intensive care unit (ICU). Next, we classified outpatient visits into 39 diagnosis categories and analysed the change in proportion of high-intensity visits versus the change in number of services. Finally, we quantified the extent to which trends in high-intensity billing are explained by changes in patient demographics and services provided in the ED using multivariable modelling.ResultsHigh-intensity visits grew from 45.8% of 671 103 visits in 2006 to 57.8% of 629 010 visits in 2012 (2.0% absolute increase per year; 95% CI 1.97% to 2.03%) as did the mean number of services provided for admitted (1.28 to 1.41; +0.02 increase in procedures per year; 95% CI 0.018 to 0.021) and discharged ED patients (7.1 to 8.6; +0.25 increase in services per year; 95% CI 0.245 to 0.255). There was a reduction in hospital admission rate from 40.1% to 35.9% (−0.68% per year; 95% CI −0.71% to −0.65%; P<0.001), while the ICU rate of admission rose from 11.7% to 12.3% (+0.11% per year; 95% CI 0.09% to 0.12%; P<0.001). When we stratified by diagnosis category, there was a moderate correlation between change in visits billed as high intensity and the change in mean number of services provided per visit (r=0.38; 95% CI 0.07 to 0.63). Trends in patient characteristics and services provided accounted moderately for the trend in practice intensity for outpatient visits (pseudo R2of 0.47) but very little for inpatient visits (0.051) and visits overall (0.148).ConclusionsIncreases in services provided in the ED moderately account for the trends in billing for high-intensity emergency care for outpatient visits.

Download Full-text

Patient mortality after surgery on the surgeon’s birthday: observational study

BMJ ◽

10.1136/bmj.m4381 ◽

2020 ◽

pp. m4381

Author(s):

Hirotaka Kato ◽

Anupam B Jena ◽

Yusuke Tsugawa

Keyword(s):

Mortality Rate ◽

Observational Study ◽

Life Events ◽

Fixed Effects ◽

Severity Of Illness ◽

Patient Characteristics ◽

Critical Access Hospitals ◽

Fee For Service ◽

Medicare Beneficiaries ◽

Study Analysis

Abstract Objective To determine whether patient mortality after surgery differs between surgeries performed on surgeons’ birthdays compared with other days of the year. Design Retrospective observational study. Setting US acute care and critical access hospitals. Participants 100% fee-for-service Medicare beneficiaries aged 65 to 99 years who underwent one of 17 common emergency surgical procedures in 2011-14. Main outcome measures Patient postoperative 30 day mortality, defined as death within 30 days after surgery, with adjustment for patient characteristics and surgeon fixed effects. Results 980 876 procedures performed by 47 489 surgeons were analyzed. 2064 (0.2%) of the procedures were performed on surgeons’ birthdays. Patient characteristics, including severity of illness, were similar between patients who underwent surgery on a surgeon’s birthday and those who underwent surgery on other days. The overall unadjusted 30 day mortality on the operating surgeon’s birthday was 7.0% (145/2064) and that on other days was 5.6% (54 824/978 812). After adjusting for patient characteristics and surgeon fixed effects (effectively comparing outcomes of patients treated by the same surgeon on different days), patients who underwent surgery on a surgeon’s birthday exhibited higher mortality compared with patients who underwent surgery on other days (adjusted mortality rate, 6.9% v 5.6%; adjusted difference 1.3%, 95% confidence interval 0.1% to 2.5%; P=0.03). Event study analysis of patient mortality by day of surgery relative to a surgeon’s birthday found similar results. Conclusions Among Medicare beneficiaries who underwent common emergency surgeries, those who received surgery on the surgeon’s birthday experienced higher mortality compared with patients who underwent surgery on other days. These findings suggest that surgeons might be distracted by life events that are not directly related to work.

Download Full-text

Group sequential designs for stepped-wedge cluster randomised trials

Clinical Trials ◽

10.1177/1740774517716937 ◽

2017 ◽

Vol 14 (5) ◽

pp. 507-517 ◽

Cited By ~ 5

Author(s):

Michael J Grayling ◽

James MS Wason ◽

Adrian P Mander

Keyword(s):

Randomised Trial ◽

Cluster Randomised Trial ◽

Type I ◽

Randomised Trials ◽

Early Stopping ◽

Stepped Wedge ◽

Group Sequential ◽

Interim Analyses ◽

Cluster Randomised Trials ◽

Cluster Randomised

Background/Aims: The stepped-wedge cluster randomised trial design has received substantial attention in recent years. Although various extensions to the original design have been proposed, no guidance is available on the design of stepped-wedge cluster randomised trials with interim analyses. In an individually randomised trial setting, group sequential methods can provide notable efficiency gains and ethical benefits. We address this by discussing how established group sequential methodology can be adapted for stepped-wedge designs. Methods: Utilising the error spending approach to group sequential trial design, we detail the assumptions required for the determination of stepped-wedge cluster randomised trials with interim analyses. We consider early stopping for efficacy, futility, or efficacy and futility. We describe first how this can be done for any specified linear mixed model for data analysis. We then focus on one particular commonly utilised model and, using a recently completed stepped-wedge cluster randomised trial, compare the performance of several designs with interim analyses to the classical stepped-wedge design. Finally, the performance of a quantile substitution procedure for dealing with the case of unknown variance is explored. Results: We demonstrate that the incorporation of early stopping in stepped-wedge cluster randomised trial designs could reduce the expected sample size under the null and alternative hypotheses by up to 31% and 22%, respectively, with no cost to the trial’s type-I and type-II error rates. The use of restricted error maximum likelihood estimation was found to be more important than quantile substitution for controlling the type-I error rate. Conclusion: The addition of interim analyses into stepped-wedge cluster randomised trials could help guard against time-consuming trials conducted on poor performing treatments and also help expedite the implementation of efficacious treatments. In future, trialists should consider incorporating early stopping of some kind into stepped-wedge cluster randomised trials according to the needs of the particular trial.

Download Full-text

Re: Tadalafil 5 mg Once Daily for the Treatment of Erectile Dysfunction during a 6-Month Observational Study (EDATE): Impact of Patient Characteristics and Comorbidities

The Journal of Urology ◽

10.1016/j.juro.2016.04.037 ◽

2016 ◽

Vol 196 (2) ◽

pp. 515-516

Author(s):

Allen D. Seftel

Keyword(s):

Erectile Dysfunction ◽

Observational Study ◽

Patient Characteristics ◽

Once Daily

Download Full-text

Gender as a Prognostic Factor in CLL. Biological Pointers to the Improved Outcome of Women.

Blood ◽

10.1182/blood.v104.11.957.957 ◽

2004 ◽

Vol 104 (11) ◽

pp. 957-957 ◽

Cited By ~ 4

Author(s):

Daniel Catovsky ◽

Estella Matutes ◽

Alison Morilla ◽

Anna Burford ◽

Vasantha Brito-Babapulle ◽

...

Keyword(s):

Cox Regression ◽

Randomised Trial ◽

Lower Proportion ◽

Response To Therapy ◽

Response To Treatment ◽

P Value ◽

Fish Analysis ◽

Randomised Trials ◽

Cox Regression Analysis ◽

Mutational Status

Abstract Gender is not widely regarded as a prognostic indicator in CLL. However, the combined data from three MRC randomised trials, CLL1, 2 and 3, and two observational studies for patients with Binet stage A, CLL2A and 3A, over a period of 20 years (1978-1998) totalling 2370 patients, showed a significant survival advantage for women (2p<0.0001). Cox regression analysis of the three randomised trials showed that gender, age, stage and response to therapy were independent prognostic variables. The response to treatment for women was also better than for men receiving the same therapies. The LRF CLL4 randomised trial, which started in 1999 shows the same trend. Preliminary results in 444 patients Binet stages A progressive, B and C showed the same CR/Nod.PR for both sexes (43.5%) but a higher PR rate in women (45%) vs men (36.5%) and a lower proportion of women non-responders to first line therapy (11.5% vs 20%). A number of laboratory investigations in CLL4, which included FISH analysis with five probes, VH mutational status, CD38 and ZAP-70 expression by flow cytometry, showed differences between the sexes, which were significant for 17p and 11q deletions combined and CD38, always in favour of women, as shown in Tables 1 and 2. The clinical and laboratory results suggest that CLL is biologically more benign in women. Women have a lower incidence of CLL, an overall higher incidence of stage A (41.7%) than men (27.3%) in CLL 1, 2, 2A, 3 and 3A and respond better to treatment in all the trials. These differences may be underlined by a higher proportion of 13q del as sole abnormality, a lower proportion of 17p (p53 locus) and 11q deletions and lower levels of CD38. Data on VH mutations and ZAP-70 point in the same direction but the number of cases studied is still small. An additional factor that may play a role in the better outcome for women relates to the effect of oestrogen derivatives which are known to target selectively superoxide dismutase and induce cell kill (Huang et al, Nature407, 390, 2000). Table 1: FISH analysis by gender (Dohner hierarchical model) Abnormality 17p del 11q del Trisomy 12 13q del Others * = Combined p value < 0.05 (Chi-Square test) Men (286) 12% 19% 10% 34% 25% Women (94) 7% 13% 11% 44% 25% p value * * 0.052 Table 2: Other biological markers CD38 negative (<30%) VH mutated ZAP-70 negative Men 173/335 (52% ) 51/149 (34%) 94/192 (49%) Women 68/100 (68%) 23/52 (44%) 42/68 (62 %) p value <0.005 NS 0.07

Download Full-text

HD12 Randomised Trial Comparing 8 Dose-Escalated Cycles of BEACOPP with 4 Escalated and 4 Baseline Cycles in Patients with Advanced Stage Hodgkin Lymphoma (HL): An Analysis of the German Hodgkin Lymphoma Study Group (GHSG), University of Cologne, D-50924 Cologne, Germany.

Blood ◽

10.1182/blood.v108.11.99.99 ◽

2006 ◽

Vol 108 (11) ◽

pp. 99-99 ◽

Cited By ~ 4

Author(s):

Andreas Engert ◽

Jeremy Franklin ◽

Rolf-Peter Mueller ◽

Hans-Theodor Eich ◽

Axel Gossmann ◽

...

Keyword(s):

Factorial Design ◽

Hodgkin Lymphoma ◽

Sequential Analysis ◽

Residual Disease ◽

Randomised Trial ◽

Final Analysis ◽

Patient Characteristics ◽

Preceding Trial ◽

Diagnostic Panel ◽

Lymphoma Study Group

Abstract The preceding trial HD9 had shown a significant superiority for escalated BEACOPP (EB) for failure-free survival (FFTF) and overall survival (OS) over COPP/ABVD or baseline BEACOPP (BB) (each 8 cycles). HD12 aimed to de-escalate chemotherapy by comparing 8 cycles EB with 4 cycles EB plus 4 cycles BB, with or without consolidatory 30 Gy radiation (RT) to initial bulky and residual disease. A total of 1661 patients aged 16–65 with HL in stage IIB (large mediastinal mass and/or E-lesions) or stage III-IV were randomised from 9/1999-1/2003 according to a factorial design between: 8EB + RT, 8EB no RT, 4EB+4BB + RT, 4EB+4BB no RT. A central diagnostic panel reviewed CT images after completion of chemotherapy independent of randomisation arm, and prescribed RT in selected cases. In the fifth analysis of HD12 (5/2006), 94% of all 1593 eligible randomised patients were evaluable, including 1498 for the chemotherapy comparison and 1449 for the radiotherapy comparison. Patient characteristics were very similar for all 4 arms. Radiation was given in arms with planned RT in 65%, in arms with no planned RT in 10%. 93% of all patients reached a complete remission, 2.2% suffered early progression, 4.6% relapsed and 8.2% died with a median folllow-up time of 48 months. Death due to acute toxicity was 3.2% (sepsis, cardiac, pulmonary, infection), with 21 such deaths in the 8EB arms and 27 in the 4EB+4BB arms. Secondary neoplasias were observed in 55 patients (3.7%): 14 AML/MDS, 12 NHL and 29 solid tumors/others. Four-year Kaplan-Meier estimates for arm pairs according to the factorial design were: 4 year rates, %(95% CI) 8EB vs. 4EB+4BB with RT vs. without RT FFTF 88(86–91) 86(83–89) 91(89–93) 88(86–91) OS 93(91–95) 91(89–93) No significant differences were observed when comparing treatments according to the sequential analysis plan (N.B. 10% received RT in arms ‘without RT’). The final analysis, scheduled when 80% of patients have 5 years follow-up, must be awaited before final conclusions can be drawn.

Download Full-text

The challenges of research on new ways to support recovery

Mental Health and Social Inclusion ◽

10.1108/mhsi-08-2014-0027 ◽

2014 ◽

Vol 18 (4) ◽

pp. 169-175

Author(s):

Sue Holttum

Keyword(s):

Psychotic Disorders ◽

Controlled Trial ◽

Randomised Trial ◽

Pilot Trial ◽

Real Life ◽

Randomised Trials ◽

Therapeutic Approaches ◽

Positive Psychotherapy ◽

Content Type ◽

New Therapeutic Approaches

Purpose – The randomised controlled trial, though highly valued, has been criticised as not helping to understand how results occur: Real-life complexity is not captured, i.e. what actually happens at trial sites (rather than what was intended). The purpose of this paper is to summarise and comment on two 2014 research papers addressing this challenge of randomised trials – concerning new therapeutic approaches for people diagnosed with psychotic disorders. Design/methodology/approach – One paper is about what staff thought when adopting a new recovery-focused approach in two mental health services as part of a randomised trial. The other is the plan for a small pilot trial of a new treatment for psychosis called positive psychotherapy. It describes how the researchers planned to study the detail of what happens in their small trial, to help them improve the design of a future, larger trial. Findings – The first paper recommends avoiding services undergoing too many changes and ensuring managers will visibly support the project. When training staff in a new approach, trainers should recognise staff's existing knowledge and skills and use practical methods like role-play. In the second paper, the plan for the small positive psychotherapy trial seems detailed enough to explain what really happens, except in one area: looking at how clinicians actually select service users for the trial. Originality/value – These papers concern pioneering therapeutic approaches in psychosis. With randomised trials highly influential, both these papers recognise their potential problems, and seem to represent good attempts to understand what really happens.

Download Full-text