I am no expert in these matters, but I would think that being one of the original authors of this systematic review/meta-analysis
(see Sartorius, B., Sartorius, K., Aldous, C., Madiba, T. E., Stefan, C., & Noakes, T. (2016). Carbohydrate intake, obesity, metabolic syndrome and cancer risk? A two-part systematic review and meta-analysis protocol to estimate attributability. BMJ open, 6(1), e009301 )
is a conflict of interest when it comes to serve as a reviewer for the very same systematic review/meta-analysis. I feel that withdrawing authorship from the manuscript reporting the results of the systematic review/meta-analysis is not enough to eliminate this conflict of interest. In the documents available via the article info on the BMJ Open website, it seems that Professor Tim Noakes has not declared this potential conflict of interest.
I would like the Editors of BMJ Open to provide some clarifications on this matter.
This paper is a very valuable addition to the literature on clinical trial transparency. It illustrates once again that self-regulation by the medical research sector does not work.
One limitation is the data provided in the supplementary appendix. The authors could have listed the full information gathered on each trial, including trial number (where available), name of PI, and sponsor name. As the recent STAT investigation into unreported trials (https://www.statnews.com/2018/01/09/clinical-trials-reporting-nih/) has shown, making performance transparent in and of itself can drive the subsequent adoption of best practices, which is in the interests of patients and the research community alike.
Even if confidentiality agreements precluded the identification of trials, it may have been possible to include more granular data without enabling the re-identification of specific trials. For example, many countries and funders have laws, policies and regulations that make prospective trial registration compulsory (https://docs.wixstatic.com/ugd/01f35d_def0082121a648529220e1d56df4b50a.pdf).
Knowing in which countries (rather than aggregated global regions) the unregistered and retrospectively registered trials were conducted, which funders (rather than aggregated funder types) funded them,...
This paper is a very valuable addition to the literature on clinical trial transparency. It illustrates once again that self-regulation by the medical research sector does not work.
One limitation is the data provided in the supplementary appendix. The authors could have listed the full information gathered on each trial, including trial number (where available), name of PI, and sponsor name. As the recent STAT investigation into unreported trials (https://www.statnews.com/2018/01/09/clinical-trials-reporting-nih/) has shown, making performance transparent in and of itself can drive the subsequent adoption of best practices, which is in the interests of patients and the research community alike.
Even if confidentiality agreements precluded the identification of trials, it may have been possible to include more granular data without enabling the re-identification of specific trials. For example, many countries and funders have laws, policies and regulations that make prospective trial registration compulsory (https://docs.wixstatic.com/ugd/01f35d_def0082121a648529220e1d56df4b50a.pdf).
Knowing in which countries (rather than aggregated global regions) the unregistered and retrospectively registered trials were conducted, which funders (rather than aggregated funder types) funded them, and which individual journals published them, would immediately flag up systems that are currently not working as intended.
This would provide a useful entry point for other researchers to examine why current safeguards are not working in these specific countries, funding agencies, and journals. It would also alert those working in the responsible entities that their current systems are not working as intended, and thus enable them to take corrective action.
A fascinating review showing both over and under use of tests. One might conclude that we're not very good at following guidelines. Another view would be that guideline developers have failed to develop implementation plans and then track whether the guidance is being followed. As a General Practitioner I wonder how I'd perform if tested on the detail of the guidelines that were relevant to my practice. One factor that will be driving under / over testing will simply be ignorance. Only this morning I had to remind a colleague that the local antibiotic of choice for community pneumonia was amoxicillin not doxycyline. We need to recognise that some aspects of medicine are now too complex to rely on the practitioner's memory and consider developing minimally intrusive software which makes it easier to do the right thing.
We read the article titled "Screening for chronic kidney disease in a community-based diabetes cohort in rural Guatemala: a cross-sectional study" published in BMJ Open. The article summoned our interest as your research work is very much relevant to the operational research in the field of diabetes. The studies in the past have consistently reported, early screening for diabetes complications, risk stratification and adhering to management guidelines are the mainstay to reduce premature mortality. As estimates show almost 20% of all ESRD cases may be attributed to diabetes alone, the targeted screening for CKD among diabetes patients becomes imperative.1
Though renal complications of diabetes are known to be common among persons with diabetes, there are no global estimates of burden and yield of targeted screening. Most of the data regarding the burden of CKD among diabetics is from isolated studies mostly from high-income countries.2 These studies have shown varied estimates based on the criteria’s, standards and techniques used to determine CKD in each of these studies. Also, a majority of the studies were based in tertiary care facilities and might have failed to report the real burden of the disease of interest in the risk group. In its latest guidelines, American Diabetes Association (ADA) recommended screening for Diabetes Kidney Disease through assessment of albuminuria and estimated glomerular filtration rate in all patients...
We read the article titled "Screening for chronic kidney disease in a community-based diabetes cohort in rural Guatemala: a cross-sectional study" published in BMJ Open. The article summoned our interest as your research work is very much relevant to the operational research in the field of diabetes. The studies in the past have consistently reported, early screening for diabetes complications, risk stratification and adhering to management guidelines are the mainstay to reduce premature mortality. As estimates show almost 20% of all ESRD cases may be attributed to diabetes alone, the targeted screening for CKD among diabetes patients becomes imperative.1
Though renal complications of diabetes are known to be common among persons with diabetes, there are no global estimates of burden and yield of targeted screening. Most of the data regarding the burden of CKD among diabetics is from isolated studies mostly from high-income countries.2 These studies have shown varied estimates based on the criteria’s, standards and techniques used to determine CKD in each of these studies. Also, a majority of the studies were based in tertiary care facilities and might have failed to report the real burden of the disease of interest in the risk group. In its latest guidelines, American Diabetes Association (ADA) recommended screening for Diabetes Kidney Disease through assessment of albuminuria and estimated glomerular filtration rate in all patients with type 2 diabetes. The 24-hour urine estimation of albumin was considered as the gold standard to detect albuminuria. However, as it is not feasible, ADA recommends the use of urinary albumin-to-creatinine ratio (UACR) in a random spot urine collection to screen for albuminuria.3 In this context, your study conducted among community cohort from LMIC with globally accepted criteria for the classification of CKD has addressed many concerns posed by the research works in the past.
However few results presented in the article needs to be interpreted with caution. It is reported that majority (65.2%) of the participants had ACR values in the normal range (<30mg/g). This estimation may be an underestimate as almost one-third of the participants were already on ACE inhibitors. Irrespective of the reason for initiating on ACE inhibitors, these drugs might have had reduced the albuminuria among a substantial proportion of the users. This would have led to normal albuminuria values among the participants with ACE inhibitors use, or else who might have had higher urine albumin values. This would have also contributed to errors in estimates of the risk stratification based on albumin values and eGFR. This may also be the potential reason for high normoalbuminuria among those reduced eGFR in the study cohort. The exploratory analysis of CKD risk map with KDIGO designations for the subgroups stratified based on ACE inhibitors use would have added to a better understanding of the study results. Though with these limitations, the study results provide great insights on the burden of CKD among treated diabetes patients in a programmatic setting.
References:
1. Chen J. Diabetic Nephropathy: Scope of the Problem. In: Edgar V L, Vecihi B, editors. Diabetes Kidney Disease [Internet]. New York: Springer; 2014. p. 9–14. Available from: http://www.springer.com/gp/book/9781493907922
2. Roglic G, World Health Organization, editors. Global report on diabetes. Geneva, Switzerland: WHO; 2016. 86 p.
3. American Diabetes Association. Standards of Medical Care in Diabetes -2017. Diabetes Care 2017;40(Supplement 1):S1–S2.
Ramsay points out that the UK’s ageing population was predicted to lead to an increase or slower rate of decline in mortality rates. However, our findings suggest that the changes in mortality rate are robust to an ageing population. In Supplementary Table S4, we show that most of the age groups above 60+ have significantly higher-than-expected mortality rates in 2012, 2013, and 2014. In particular, the mortality rates for the 85+ age group is consistently higher than expected for all years. If population mortality rates were expected to spike mostly as the result of an increasing proportion of very old people, we would not see such big spikes within these older age groups.
As noted by Black, our use of the 1976 European standard population (ESP) reference instead of the 2013 ESP for age-standardising death rates (ASDR) is the source of the discrepancy between our ASDR and more recently published ONS ASDR.
We calculated ASDr de novo using raw mortality counts and population data since ASDR broken down by sex and specific to England (not England and Wales) were unavailable at the time of data collection and initial data processing (mid-2014). To do this, we used the 1976 ESP data since to our knowledge, the 2013 ESP data were not available at that time.
To demonstrate that the source of standardisation did not affect our conclusion, we re-processed the data using 2013 ESP data and re-ran the time-trend analyses. Despite this difference, we again found that the spending constraints of 2010/11 were linked with a significant increase in ASDR. Comparing the actual and predicted ASDR revealed 12,111 higher than expected number of deaths in 2012 (95% CI 4,912 to 19,309), 23,311 (95% CI 15,994 to 30,628) in 2013, and 20,351 (95% CI 12,865 to 27,838) in 2014 (see Figure 1: http://blogs.bmj.com/bmjopen/files/2018/01/BMJ-Open-Response-Figure-1.jpg). These numbers are all within the corresponding confidence intervals of the results using the 1976 ESP-standardised data.
Furthermore, it is worth noting that our findings that life expectancy was si...
As noted by Black, our use of the 1976 European standard population (ESP) reference instead of the 2013 ESP for age-standardising death rates (ASDR) is the source of the discrepancy between our ASDR and more recently published ONS ASDR.
We calculated ASDr de novo using raw mortality counts and population data since ASDR broken down by sex and specific to England (not England and Wales) were unavailable at the time of data collection and initial data processing (mid-2014). To do this, we used the 1976 ESP data since to our knowledge, the 2013 ESP data were not available at that time.
To demonstrate that the source of standardisation did not affect our conclusion, we re-processed the data using 2013 ESP data and re-ran the time-trend analyses. Despite this difference, we again found that the spending constraints of 2010/11 were linked with a significant increase in ASDR. Comparing the actual and predicted ASDR revealed 12,111 higher than expected number of deaths in 2012 (95% CI 4,912 to 19,309), 23,311 (95% CI 15,994 to 30,628) in 2013, and 20,351 (95% CI 12,865 to 27,838) in 2014 (see Figure 1: http://blogs.bmj.com/bmjopen/files/2018/01/BMJ-Open-Response-Figure-1.jpg). These numbers are all within the corresponding confidence intervals of the results using the 1976 ESP-standardised data.
Furthermore, it is worth noting that our findings that life expectancy was significantly lower than expected and the rate of potential years of life lost was significantly higher than expected following the health and social care spending constraints provide further evidence for our conclusion.
The type of vertebral fractures object of the study of Smits et al. is defined as A2-C by AO [1]. The classification of AO of a thoracolumbar fracture includes three types of fractures: type A, type B and type C. Accordingly, type A has five subtypes:
А0 – minor, nonstructural fractures;
А1 – wedge-compression;
А2 – split;
А3 – incomplete burst;
А4 – complete burst .
Thoracolumbar fractures of type A2-C are not present in the AO classification [2].
The article states that forty-six patients enrolled in the study were between 18 to 65 years of age but did not refer to the distribution of patients by age and gender. The bone quality will be different in one 18-year-old patient and a 65-year-old patient respectively [3]. Bone quality is important for the attachment of pedicular screw synthesis and it can be compromised in osteoporotic patients [4,5]. The use of three-point hyperextension brace in these patients postoperatively would be advisable [6].
Abbreviation: AO – Müller AO Classification of fractures.
References
1. Smits AJ, Deunk J, Stadhouder A, Altena MC, Kempen DHR, Bloemers FW. Is postoperative bracing after pedicle screw fixation of spine fractures necessary? Study protocol of the ORNOT study: a randomised controlled. http://bmjopen.bmj.com/content/8/1/e019596
2. Spinal Fractures Classification System - AO: Tools....
The type of vertebral fractures object of the study of Smits et al. is defined as A2-C by AO [1]. The classification of AO of a thoracolumbar fracture includes three types of fractures: type A, type B and type C. Accordingly, type A has five subtypes:
А0 – minor, nonstructural fractures;
А1 – wedge-compression;
А2 – split;
А3 – incomplete burst;
А4 – complete burst .
Thoracolumbar fractures of type A2-C are not present in the AO classification [2].
The article states that forty-six patients enrolled in the study were between 18 to 65 years of age but did not refer to the distribution of patients by age and gender. The bone quality will be different in one 18-year-old patient and a 65-year-old patient respectively [3]. Bone quality is important for the attachment of pedicular screw synthesis and it can be compromised in osteoporotic patients [4,5]. The use of three-point hyperextension brace in these patients postoperatively would be advisable [6].
Abbreviation: AO – Müller AO Classification of fractures.
References
1. Smits AJ, Deunk J, Stadhouder A, Altena MC, Kempen DHR, Bloemers FW. Is postoperative bracing after pedicle screw fixation of spine fractures necessary? Study protocol of the ORNOT study: a randomised controlled. http://bmjopen.bmj.com/content/8/1/e019596
2. Spinal Fractures Classification System - AO: Tools. www.aosla.com.br/ fractures/classification.pdf – accessible to17.01.2018.
3. Fonseca H, Moreira-Gonçalves D, Coriolano HJ, Duarte JA. Bone quality: the determinants of bone strength and fragility. Sports Med. 2014 Jan;44(1):37-53.
4. Halverson TL, Kelley LA, Thomas KA, Whitecloud TS, Cook SD. Efects of bone mineral density on pedicle screw fixation. Spine 1994; 9(21):2415–20.
5. Cook SD, Salkeld SL, Stanley T, Faciane A, Miller SD.Biomechanical study of pedicle screw fixation in severely osteoporotic bone. Spine J. 2004 Jul-Aug;4(4):402-8.
6. Longo UG, Loppini M, Denaro L, Maffulli N, Denaro V. Osteoporotic vertebral fractures: current concepts of conservative care. Br Med Bull. 2012 Jun;102:171-89.
Kumar raises concerns about the interpretation of our study examining the cost savings of different repeat prescription durations, and argues monthly dispensing should be maintained. We thank Mr Kumar for his comments and interest in our article and respond below.
In response to Kumar's first point, our study suggests that with shorter prescription lengths, the cost savings due to reductions in waste are more than offset by the additional dispensing fees; these fees can therefore not be considered "nominal". White1 further highlights the non-nominal impact of additional dispensing fees due to short prescriptions lengths, estimating an additional cost of £700 million if all prescription items issued by the NHS in England were issued as 28-day repeat-dispensing items.
We are also not aware of any evidence that increased frequency of contact with the community pharmacist (or indeed “indirect contact" with the GP authorising a prescription) due to shorter prescription lengths impacts in a valuable way in terms of patient outcomes as implied, either from a clinical or cost effectiveness perspective. Work by Elliott et al.2 has shown the pharmacy-based New Medicines Service to be cost-effective at improving adherence. However, un-targeted clinical pharmacy services, more generally, lack conclusive evidence of their effect on medication adherence and prescription appropriateness.3 Indeed, in many cases, the contact between patient and pharmacist...
Kumar raises concerns about the interpretation of our study examining the cost savings of different repeat prescription durations, and argues monthly dispensing should be maintained. We thank Mr Kumar for his comments and interest in our article and respond below.
In response to Kumar's first point, our study suggests that with shorter prescription lengths, the cost savings due to reductions in waste are more than offset by the additional dispensing fees; these fees can therefore not be considered "nominal". White1 further highlights the non-nominal impact of additional dispensing fees due to short prescriptions lengths, estimating an additional cost of £700 million if all prescription items issued by the NHS in England were issued as 28-day repeat-dispensing items.
We are also not aware of any evidence that increased frequency of contact with the community pharmacist (or indeed “indirect contact" with the GP authorising a prescription) due to shorter prescription lengths impacts in a valuable way in terms of patient outcomes as implied, either from a clinical or cost effectiveness perspective. Work by Elliott et al.2 has shown the pharmacy-based New Medicines Service to be cost-effective at improving adherence. However, un-targeted clinical pharmacy services, more generally, lack conclusive evidence of their effect on medication adherence and prescription appropriateness.3 Indeed, in many cases, the contact between patient and pharmacist will be extremely brief, and in the case of individuals on long-standing therapy the opportunity for meaningful clinical input will be limited. The fact that patients report greater inconvenience due to collecting scripts more frequently also suggests that patients do not perceive any particular added value.4
The electronic prescribing systems referred to in Kumar's second point were indeed not taken into account by our analysis. However, a substantial proportion of prescriptions continue to be dispensed using the standard repeat prescribing system: according to Electronic Prescription Service (EPS) National Statistics for England, despite over 90% of GP practices and pharmacies being registered, utilisation of EPS is only 59% (percentage of all items claimed, which were claimed via EPS) and electronic Repeat Dispensing (eRD) utilisation even lower at 8% (percentage of all items claimed, which were prescribed via eRD).5 Even if utilisation rates can be increased it is estimated that only 80% of repeat prescriptions could be replaced with eRD; suggesting in an ideal world, that 20% of repeat prescription would still need to be issued non-electronically.6
Unfortunately, differentiating EPS from standard repeat prescribing is not readily done in CPRD. Nonetheless, the greater dispensing fees incurred due to greater frequency of prescribing are still considerably higher than the resulting savings in terms of waste, even when prescriber time is not considered. The two examples we give for antidepressants and two statins are still savings of £104 million and £62 million respectively even if prescriber time savings are removed.
Finally, we would reemphasise our conclusion that longer repeat prescription lengths should be considered for common chronic conditions “where clinically appropriate”. Those with unstable or uncontrolled conditions would clearly be excluded but in cases where patients are well managed, we recommend GPs consider a longer repeat interval.
References
1. White KG. UK interventions to control medicines wastage: a critical review. The International journal of pharmacy practice 2010;18(3):131-40.
2. Elliott RA, Tanajewski L, Gkountouras G, et al. Cost Effectiveness of Support for People Starting a New Medication for a Long-Term Condition Through Community Pharmacies: An Economic Evaluation of the New Medicine Service (NMS) Compared with Normal Practice. PharmacoEconomics 2017;35(12):1237-55.
3. Rotta I, Salgado TM, Silva ML, et al. Effectiveness of clinical pharmacy services: an overview of systematic reviews (2000-2010). International journal of clinical pharmacy 2015;37(5):687-97.
4. Miani C, Martin A, Exley J, et al. Clinical effectiveness and cost-effectiveness of issuing longer versus shorter duration (3-month vs. 28-day) prescriptions in patients with chronic conditions: systematic review and economic modelling. Health technology assessment 2017;21(78):1-128.
Thank you for your suggestion. We agree with you that parental autistic disorders might increase the risk of ASD in children. In our study, we set all parental psychiatric disorders (ICD-10 codes F00-F99) including autism disorders (ICD-10 codes F84.0 F84.1, F84.5, F84.8 and F84.9) as potential confounding factors, and we adjusted for mother psychiatric disorders (including ASD) in model 1, and father psychiatric disorders (including ASD) in model 2. Besides, we also conducted sibling-matched analyses to control for shared family-related confounding factors like genetic liability for neuropsychiatric conditions.
With interest we read the article of Wiegerinck et al. on intrapartum and neonatal mortality among low-risk women in midwife-led versus obstetrician-led care based on case notes and data from the Dutch national perinatal database (PRN, nowadays called ‘Perined’). In a recent commentary we addressed some important pitfalls associated with the use of register-based data.1 Although the authors acknowledge the methodological challenges of their study, and the limitations of the Perined database, it is important to specify these issues in more detail. Secondly, we want to challenge the interpretation of the findings.
In this study clinical information regarding the women and babies in the numerators and denominators for intrapartum and neonatal mortality rates were derived from different sources. For the numerators information on women and their babies who died was taken from case notes. For the denominators information was taken from the Perined database, in which information on risk factors is often missing. The authors show in their supplementary file that in a sample of 100 women who started labour in obstetrician-led care no information was recorded for 13% of women on one or more of the eight risk factors which were defined as exclusion criteria in this study. Most women with risk factors will have been removed from the numerator, because information was available in their case notes. On the other hand, an unknown number of women with risk factors were not removed f...
With interest we read the article of Wiegerinck et al. on intrapartum and neonatal mortality among low-risk women in midwife-led versus obstetrician-led care based on case notes and data from the Dutch national perinatal database (PRN, nowadays called ‘Perined’). In a recent commentary we addressed some important pitfalls associated with the use of register-based data.1 Although the authors acknowledge the methodological challenges of their study, and the limitations of the Perined database, it is important to specify these issues in more detail. Secondly, we want to challenge the interpretation of the findings.
In this study clinical information regarding the women and babies in the numerators and denominators for intrapartum and neonatal mortality rates were derived from different sources. For the numerators information on women and their babies who died was taken from case notes. For the denominators information was taken from the Perined database, in which information on risk factors is often missing. The authors show in their supplementary file that in a sample of 100 women who started labour in obstetrician-led care no information was recorded for 13% of women on one or more of the eight risk factors which were defined as exclusion criteria in this study. Most women with risk factors will have been removed from the numerator, because information was available in their case notes. On the other hand, an unknown number of women with risk factors were not removed from the denominator, because this information was missing in the Perined database. Therefore, the real mortality rate might be higher than what has been reported. Moreover, 12% of the 100 women in obstetrician-led care, according to the Perined database, had started labour in midwife-led care which indicates another potential problem with the size of the denominator. The authors controlled for differences in characteristics between the women in midwife-led and obstetrician-led care using a propensity score matched (PSM) pairs analysis. Unfortunately, this does not solve the problems with the data as no correction can be made for risk factors for which information is not available or incorrect.
We conclude that the pitfalls regarding the definition of the study population (numerator and denominator) and the composition of the comparison groups were not taken into account sufficiently in this article. These methodological issues add to the limitations of this study.
In addition to these limitations, the authors hypothesize that the continuous presence of an obstetric care provider and continuous CTG monitoring are important determinants of the non-significant difference found in intrapartum and neonatal mortality rates. This hypothesis is contrary to the international concern about high intervention rates without evidence of benefits.2,3 In particular, routine continuous electronic fetal monitoring and the routine involvement of obstetricians in the care of women with a normal course of pregnancy have not been recommended for improvement of perinatal results.3
Continuous electronic fetal monitoring has not been shown to lead to better long term neonatal outcomes and is associated with a higher rate of caesarean section.4 This study showed a 3.5 fold lower CS rate among women who started labour in midwife-led care. In addition, lower rates of instrumental vaginal birth, epidural anaesthesia, episiotomy, manual removal of placenta and postpartum haemorrhage were found among these women. These findings, although statistically highly significant, should also be interpreted with caution, in view of the limitations of the data. Nevertheless, a discussion of the adverse consequences of caesarean sections and other interventions would have been appropriate. CS is associated, for example, with higher rates of severe adverse maternal morbidity5 and higher rates of unexplained stillbirth6 or uterine rupture during the next pregnancy.7
It is important to balance the use of life-saving obstetric interventions among women who need them, while avoiding adverse effects of inappropriate use. Research findings can help professionals to make evidence-based decisions about when to offer women interventions and specialist obstetric care. However, we postulated earlier that it is not possible to reliably compare low-risk women who started labour in midwife-led versus obstetrician-led care in the Netherlands on the basis of Perined data.1 This work of Wiegerinck et al did not change our opinion and underlines the need for improving the quality of the data. If data are unreliable, findings only confuse the debate on the safety of midwife-led and obstetrician-led care.
References
1. De Jonge A, etal. Pitfalls in the use of register-based data for
comparing adverse maternal and perinatal outcomes in different birth settings.
BJOG. 2017 Sep;124(10):1477-1480.
2. Renfrew MJ, etal. Midwifery and quality care: findings from a new evidence-informed framework for maternal and newborn care. Lancet. 2014 Sep 20;384(9948):1129-45.
3. Miller S, etal. Beyond too little, too late and too much, too soon: a pathway towards evidence-based,
respectful maternity care worldwide. Lancet. 2016 Oct 29;388(10056):2176-2192.
4. Alfirevic Z, etal. Continuous cardiotocography (CTG) as a form of electronic fetal monitoring (EFM) for fetal assessment during labour. Cochrane Database Syst Rev. 2017 Feb 3;2:CD006066.
5. Van Dillen J, etal. Severe acute maternal morbidity and mode of delivery in the Netherlands. Acta Obstet Gynecol Scand. 2010 Nov;89(11):1460-5.
6. Moraitis AA, etal. Previous caesarean delivery and the risk of unexplained stillbirth: retrospective cohort study and meta-analysis. BJOG. 2015 Oct;122(11):1467-74.
7. Kok N, etal. Risk of maternal and neonatal complications in subsequent pregnancy after planned
caesarean section in a first birth, compared with emergency caesarean section: a
nationwide comparative cohort study. BJOG. 2014 Jan;121(2):216-23.
I am no expert in these matters, but I would think that being one of the original authors of this systematic review/meta-analysis
(see Sartorius, B., Sartorius, K., Aldous, C., Madiba, T. E., Stefan, C., & Noakes, T. (2016). Carbohydrate intake, obesity, metabolic syndrome and cancer risk? A two-part systematic review and meta-analysis protocol to estimate attributability. BMJ open, 6(1), e009301 )
is a conflict of interest when it comes to serve as a reviewer for the very same systematic review/meta-analysis. I feel that withdrawing authorship from the manuscript reporting the results of the systematic review/meta-analysis is not enough to eliminate this conflict of interest. In the documents available via the article info on the BMJ Open website, it seems that Professor Tim Noakes has not declared this potential conflict of interest.
I would like the Editors of BMJ Open to provide some clarifications on this matter.
This paper is a very valuable addition to the literature on clinical trial transparency. It illustrates once again that self-regulation by the medical research sector does not work.
One limitation is the data provided in the supplementary appendix. The authors could have listed the full information gathered on each trial, including trial number (where available), name of PI, and sponsor name. As the recent STAT investigation into unreported trials (https://www.statnews.com/2018/01/09/clinical-trials-reporting-nih/) has shown, making performance transparent in and of itself can drive the subsequent adoption of best practices, which is in the interests of patients and the research community alike.
Even if confidentiality agreements precluded the identification of trials, it may have been possible to include more granular data without enabling the re-identification of specific trials. For example, many countries and funders have laws, policies and regulations that make prospective trial registration compulsory (https://docs.wixstatic.com/ugd/01f35d_def0082121a648529220e1d56df4b50a.pdf).
Knowing in which countries (rather than aggregated global regions) the unregistered and retrospectively registered trials were conducted, which funders (rather than aggregated funder types) funded them,...
Show MoreA fascinating review showing both over and under use of tests. One might conclude that we're not very good at following guidelines. Another view would be that guideline developers have failed to develop implementation plans and then track whether the guidance is being followed. As a General Practitioner I wonder how I'd perform if tested on the detail of the guidelines that were relevant to my practice. One factor that will be driving under / over testing will simply be ignorance. Only this morning I had to remind a colleague that the local antibiotic of choice for community pneumonia was amoxicillin not doxycyline. We need to recognise that some aspects of medicine are now too complex to rely on the practitioner's memory and consider developing minimally intrusive software which makes it easier to do the right thing.
Dear authors,
We read the article titled "Screening for chronic kidney disease in a community-based diabetes cohort in rural Guatemala: a cross-sectional study" published in BMJ Open. The article summoned our interest as your research work is very much relevant to the operational research in the field of diabetes. The studies in the past have consistently reported, early screening for diabetes complications, risk stratification and adhering to management guidelines are the mainstay to reduce premature mortality. As estimates show almost 20% of all ESRD cases may be attributed to diabetes alone, the targeted screening for CKD among diabetes patients becomes imperative.1
Though renal complications of diabetes are known to be common among persons with diabetes, there are no global estimates of burden and yield of targeted screening. Most of the data regarding the burden of CKD among diabetics is from isolated studies mostly from high-income countries.2 These studies have shown varied estimates based on the criteria’s, standards and techniques used to determine CKD in each of these studies. Also, a majority of the studies were based in tertiary care facilities and might have failed to report the real burden of the disease of interest in the risk group. In its latest guidelines, American Diabetes Association (ADA) recommended screening for Diabetes Kidney Disease through assessment of albuminuria and estimated glomerular filtration rate in all patients...
Show MoreRamsay points out that the UK’s ageing population was predicted to lead to an increase or slower rate of decline in mortality rates. However, our findings suggest that the changes in mortality rate are robust to an ageing population. In Supplementary Table S4, we show that most of the age groups above 60+ have significantly higher-than-expected mortality rates in 2012, 2013, and 2014. In particular, the mortality rates for the 85+ age group is consistently higher than expected for all years. If population mortality rates were expected to spike mostly as the result of an increasing proportion of very old people, we would not see such big spikes within these older age groups.
As noted by Black, our use of the 1976 European standard population (ESP) reference instead of the 2013 ESP for age-standardising death rates (ASDR) is the source of the discrepancy between our ASDR and more recently published ONS ASDR.
We calculated ASDr de novo using raw mortality counts and population data since ASDR broken down by sex and specific to England (not England and Wales) were unavailable at the time of data collection and initial data processing (mid-2014). To do this, we used the 1976 ESP data since to our knowledge, the 2013 ESP data were not available at that time.
To demonstrate that the source of standardisation did not affect our conclusion, we re-processed the data using 2013 ESP data and re-ran the time-trend analyses. Despite this difference, we again found that the spending constraints of 2010/11 were linked with a significant increase in ASDR. Comparing the actual and predicted ASDR revealed 12,111 higher than expected number of deaths in 2012 (95% CI 4,912 to 19,309), 23,311 (95% CI 15,994 to 30,628) in 2013, and 20,351 (95% CI 12,865 to 27,838) in 2014 (see Figure 1: http://blogs.bmj.com/bmjopen/files/2018/01/BMJ-Open-Response-Figure-1.jpg). These numbers are all within the corresponding confidence intervals of the results using the 1976 ESP-standardised data.
Furthermore, it is worth noting that our findings that life expectancy was si...
Show MoreThe type of vertebral fractures object of the study of Smits et al. is defined as A2-C by AO [1]. The classification of AO of a thoracolumbar fracture includes three types of fractures: type A, type B and type C. Accordingly, type A has five subtypes:
А0 – minor, nonstructural fractures;
А1 – wedge-compression;
А2 – split;
А3 – incomplete burst;
А4 – complete burst .
Thoracolumbar fractures of type A2-C are not present in the AO classification [2].
The article states that forty-six patients enrolled in the study were between 18 to 65 years of age but did not refer to the distribution of patients by age and gender. The bone quality will be different in one 18-year-old patient and a 65-year-old patient respectively [3]. Bone quality is important for the attachment of pedicular screw synthesis and it can be compromised in osteoporotic patients [4,5]. The use of three-point hyperextension brace in these patients postoperatively would be advisable [6].
Abbreviation: AO – Müller AO Classification of fractures.
References
1. Smits AJ, Deunk J, Stadhouder A, Altena MC, Kempen DHR, Bloemers FW. Is postoperative bracing after pedicle screw fixation of spine fractures necessary? Study protocol of the ORNOT study: a randomised controlled. http://bmjopen.bmj.com/content/8/1/e019596
Show More2. Spinal Fractures Classification System - AO: Tools....
Kumar raises concerns about the interpretation of our study examining the cost savings of different repeat prescription durations, and argues monthly dispensing should be maintained. We thank Mr Kumar for his comments and interest in our article and respond below.
In response to Kumar's first point, our study suggests that with shorter prescription lengths, the cost savings due to reductions in waste are more than offset by the additional dispensing fees; these fees can therefore not be considered "nominal". White1 further highlights the non-nominal impact of additional dispensing fees due to short prescriptions lengths, estimating an additional cost of £700 million if all prescription items issued by the NHS in England were issued as 28-day repeat-dispensing items.
We are also not aware of any evidence that increased frequency of contact with the community pharmacist (or indeed “indirect contact" with the GP authorising a prescription) due to shorter prescription lengths impacts in a valuable way in terms of patient outcomes as implied, either from a clinical or cost effectiveness perspective. Work by Elliott et al.2 has shown the pharmacy-based New Medicines Service to be cost-effective at improving adherence. However, un-targeted clinical pharmacy services, more generally, lack conclusive evidence of their effect on medication adherence and prescription appropriateness.3 Indeed, in many cases, the contact between patient and pharmacist...
Show MoreThank you for your suggestion. We agree with you that parental autistic disorders might increase the risk of ASD in children. In our study, we set all parental psychiatric disorders (ICD-10 codes F00-F99) including autism disorders (ICD-10 codes F84.0 F84.1, F84.5, F84.8 and F84.9) as potential confounding factors, and we adjusted for mother psychiatric disorders (including ASD) in model 1, and father psychiatric disorders (including ASD) in model 2. Besides, we also conducted sibling-matched analyses to control for shared family-related confounding factors like genetic liability for neuropsychiatric conditions.
With interest we read the article of Wiegerinck et al. on intrapartum and neonatal mortality among low-risk women in midwife-led versus obstetrician-led care based on case notes and data from the Dutch national perinatal database (PRN, nowadays called ‘Perined’). In a recent commentary we addressed some important pitfalls associated with the use of register-based data.1 Although the authors acknowledge the methodological challenges of their study, and the limitations of the Perined database, it is important to specify these issues in more detail. Secondly, we want to challenge the interpretation of the findings.
Show MoreIn this study clinical information regarding the women and babies in the numerators and denominators for intrapartum and neonatal mortality rates were derived from different sources. For the numerators information on women and their babies who died was taken from case notes. For the denominators information was taken from the Perined database, in which information on risk factors is often missing. The authors show in their supplementary file that in a sample of 100 women who started labour in obstetrician-led care no information was recorded for 13% of women on one or more of the eight risk factors which were defined as exclusion criteria in this study. Most women with risk factors will have been removed from the numerator, because information was available in their case notes. On the other hand, an unknown number of women with risk factors were not removed f...
Pages